You are on page 1of 16

Thin-Walled Structures 40 (2002) 167–182

www.elsevier.com/locate/tws

Mushrooming of circular tubes under dynamic


axial loading
B. Wang a,*
, G. Lu b

a
Department of Mechanical Engineering, Brunel University, Uxbridge UB8 3PH, UK
b
School of Engineering and Science, Swinburne University of Technology, Melbourne, Vic. 3122,
Australia

Abstract

Extensive studies on axial crushing of cylindrical shells have been carried out due to their
wide applications for crashworthiness. Most of the studies have been conducted under static
and low speed impact conditions to predict the deforming mechanism, resulting in typical
progressive buckling and dynamic plastic buckling, from which the energy absorption capacity
of the structural component can be predicted. Though the effect of the strain rate, inertia and
stress wave propagation can all be considered to determine the dynamic buckling modes,
the present study concentrates on a different mechanism under higher energy impact, where
mushrooming, or thickening of the shell wall is observed. Both experimental study and finite
element (FE) simulations were carried out. It was found that mushrooming is an important
feature under high speed impact but experimental results show that the high impact energy
leads to dynamic tensile fracture in the mushroomed regions and the lack of suitable models
for dynamic fracture in FE codes hinders a full understanding of the problem. © 2002 Elsevier
Science Ltd. All rights reserved.

Keywords: Dynamic buckling; Thin-walled cylindrical shell; Mushrooming; Axial crush

1. Introduction

An understanding of axial crushing of thin-walled tubes under static loading has


been well established over the last two decades [1]. The nature of the progressive
buckling process under a static mode leads to a stable deforming response in which

* Corresponding author. Tel.: +44-1895-274000; fax: +44-1895-256392.


E-mail address: bin.wang@brunel.ac.uk (B. Wang).

0263-8231/02/$ - see front matter © 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 2 6 3 - 8 2 3 1 ( 0 1 ) 0 0 0 5 7 - X
168 B. Wang, G. Lu / Thin-Walled Structures 40 (2002) 167–182

folds generally occur at one end of the tube, though these folds may appear in differ-
ent forms, such as axisymmetric, concertina and mixed mode [2]. Many industrial
applications have been developed to use its energy absorption capacity. Generally
speaking, due to the low speed of deformation, the inertia and strain rate effects are
not substantial and can be neglected. Alexander [3] was the first to provide a rigid,
perfectly plastic model of the progressive buckling process and since then the model
has been improved in several ways [4–8]. See [9] for a recent review on the topic.
When a tube is subjected to a severe dynamic axial load, the response would be
a dynamic plastic buckling one which may become quite different from a progressive
one. Wrinkling will occur along a quite large portion of the sample and folds may
also occur if the load is severe enough. Tests and modelling have been conducted
on cylindrical shells of finite length struck by a moving mass. Both inertia and strain
rate have been considered. A comprehensive review can be found in Jones’ book
[1] on both dynamic progressive and plastic buckling.
Recent developments on the axisymmetric buckling of circular cylindrical shells
have seen an interest in the effect of stress wave propagation [10–13]. It is reported
that the inertia characteristics of the shell, together with the material properties, deter-
mine particular patterns of the axial stress wave propagation, thus, causing either
dynamic plastic or dynamic progressive buckling to develop during the initial phase
of the shell response. The stress wave appears to be important in the transient process
of the buckling mechanism, but the final shape is very much determined on the
inertia properties of the shell.
Most of the experimental studies were carried out under two loading conditions:
a stationary tube struck by a moving mass G—a distal impact [14–16], and a moving
tube with an attached mass G at the end striking a rigid surface [16,17]. As the mass
G is normally comparably heavier than the tube sample itself, as a first approxi-
mation, the impact can be treated as a step load generated by the inertia of the
attached mass. For a stress wave study, a different test was carried out where a
Kolsky bar was used [11] to analyse the initiation of buckling. The problem studied
in the present investigation differs from the above in that a tube sample alone is
shot and then strikes against a hard surface. It is a proximal impact with the driving
force for deformation coming completely from the inertia force generated by the
tube sample itself. The difference in the loading condition is substantial in this case.
For instance, by ignoring the effect of stress wave, we assume that the whole tube
segment is subjected to a uniform deceleration at the moment it hits the surface.
Thus the magnitude of the axial force applied in the segment would be linear rather
than uniform, having a zero value at the free end of the tube to a maximum value
at the striking end. The higher value of the compressive force at the striking end of
the sample will force lateral deformation locally and increase the wall thickness into
a mechanism resembling a mushrooming effect. The increased wall thickness leads
to a higher resistance to buckling in that portion. As a consequence, folds occur
some distance away from the striking end. In reality, a stress wave under high-speed
impact will trigger the initial buckling modes, and then consequently interact with
the thickening mode of the shell wall, leading to a complex deformation mechanism.
The inertia and strain rate are two key issues in this study. Of interest are how
B. Wang, G. Lu / Thin-Walled Structures 40 (2002) 167–182 169

Fig. 1. Sketch of the gas gun used for the experiments.

these two factors influence the structural response, such as the location of the first
fold, final wall thickness of the tube, magnitude of the impact force at contact and
total energy absorption, in terms of tube dimensions and the striking velocity. The
complexity of the problem prevents the development of a theoretical model at this
stage, thus a combined experimental study and finite element modelling approach
was adopted. Apart from mushrooming, tests also show dynamic fracture failures at
the striking end of the tube under higher impact velocities. Due to the lack of proper
models for this failure mechanism, this phenomenon was not modelled in the finite
element (FE) analysis.

2. Experimental study

Tests were carried out with a gas gun sketched in Fig. 1. In all the tests, a tube
specimen was fired so that it would strike a rigid flat anvil. The velocity of the tube
just before impact was measured by noting the time intervals the tube passed through
three light beams positioned at fixed spacing distances apart. Aluminium and mild
steel tubes were machined in-house. The outer diameter of all the tubes was 12.55
mm in order to match the bore of the gas gun barrel. The values of the wall thick-
nesses are given in Table 1. These approximately correspond to D/t ratios of 16, 8
and 4, respectively, with L/D fixed at 5.
A total of 14 impact tests were conducted. Three or four impact velocities were

Table 1
Dimensions of test specimens. The outer diameter of all the specimens is 12.55 mm

Material Length of sample (mm) Thickness of wall (mm)

Steel 62.75 0.78


62.75 1.60
62.75 3.13
Aluminum 62.75 3.13
170 B. Wang, G. Lu / Thin-Walled Structures 40 (2002) 167–182

employed for a given specimen dimensions. Fig. 2 shows photographs of the 0.78
mm thick mild steel specimens after impact at velocities of 352, 158, 127 and 114
m/s, together with an untested specimen. It is clear that different failure modes are
present. For the low velocity, the specimen buckled axisymmetrically. But when the
velocity is high, the nominally same specimen fractured with petalling.
Fig. 3 shows similar failure patterns for the 1.60 mm thick mild steel specimens.
The impact velocities were 353, 209 and 149 m/s, respectively. A strong feature
with this specimen is that the tube wall initially thickens considerably before it
buckles into the mushrooming mode. For higher velocities, the specimen fails by
fracturing.
The 3.13 mm thick mild steel specimens are shown in Fig. 4. The corresponding
velocities are 385, 277 and 173 m/s. Again, a strong mushrooming effect is present,
for the specimen tested at 173 m/s. It is clear that the specimens have undergone
significant three dimensional plastic flow within the thickness direction, in contrast
to thin-walled tubes where plastic bending/stretching dominates.
The 3.13 mm thick aluminum specimens were also tested and their photographs
are shown in Fig. 5, for impact velocities of 361, 220 and 137 m/s, respectively.
The tests were designed to investigate the effect of material properties on the failure
modes. It is well known that aluminium is much less strain-rate sensitive. However,
the general features observed are broadly similar to the mild steel specimens.
One specimen from each thickness group shown in Figs. 2–5, with the lowest
impact velocity, was selected and sectioned. They are shown in Fig. 6. Note that
the specimen on the far left is, if viewed along the tube axis, one quarter of the tube

Fig. 2. Photograph of mild steel samples. Wall thickness=0.78 mm. Velocities from left to right are:
352 m/s, 158 m/s, 127 m/s, 114 m/s and 0 m/s (untested).
B. Wang, G. Lu / Thin-Walled Structures 40 (2002) 167–182 171

Fig. 3. Photograph of mild steel samples. Wall thickness=1.60 mm. Velocities from left to right are:
353 m/s, 209 m/s, 149 m/s and 0 m/s (untested).

Fig. 4. Photograph of mild steel samples. Wall thickness=3.13 mm. Velocities from left to right are:
385 m/s, 277 m/s, 227 m/s, 173 m/s and 0 m/s (untested).
172 B. Wang, G. Lu / Thin-Walled Structures 40 (2002) 167–182

Fig. 5. Photograph of aluminium samples. Wall thickness=3.13 mm. Velocities from left to right are
361 m/s, 220 m/s, 137 m/s and 0 m/s (untested).

Fig. 6. Photograph of section profiles of specimens from each of the thickness group corresponding to
Figs. 2–5.

specimen instead of half. It is evident that thickening of tube walls took place for
these specimens under the specified impact velocities.
B. Wang, G. Lu / Thin-Walled Structures 40 (2002) 167–182 173

3. Details of FE simulations

The tube samples selected for numerical simulations had a length of 129.6 mm,
an outer diameter of 6.4 mm, and three different wall thickness to diameter ratios
(t/D) at 1/16, 3/32 and 1/8, marked as Tube 1, Tube 2 and Tube 3, respectively.
This t/D range represents relatively thick tubes where diamond-patterned deformation
with a circumferential wave would generally not occur [17]. The material is mild
steel with a Young’s modulus of 206.8 GPa and a Poisson’s ratio of 0.3. The static
stress–strain curve shown in Fig. 7 was obtained from a pure torsion test, leading
to finite strains. This would allow large deformation to be simulated in the FE model-
ling. To model the strain rate effect, the Cowper–Symonds relationship was assumed
for the material:
q
s0
ė D 1 , s0 s0
s0
where D=40.4/s and q=5 [18].
Only axisymmetric deformation was assumed. A commercial code
ABAQUS/Explicit [19] was used and a two-dimensional model was constructed
using 2691 four-node quadric axisymmetric elements. A frictional coefficient of 0.1
was assigned between the tube and the rigid surface and 0.25 between the contacts
of the tube itself (folds), respectively. No dynamic fracture failure was introduced,
as no proper models for this phenomenon were available in the FE code. As a conse-
quence, the material was allowed to undergo excessive ductile deformation without
fracture failure. All nodes in the tube were assigned with an initial velocity before
impact ranging from 200 to 600 m/s. Fig. 8 features a typical case of Tube 1 with an
initial impact velocity of 300 m/s. The dynamic buckling response is clearly revealed.
Generally, the nominal pattern of deformation can be divided into three categories:
folds for thin tubes at relatively low speeds; mushrooming and folds at medium
speeds for all tubes, and mushrooming and wrinkles only for thick tubes at high

Fig. 7. Stress–strain curve of the tube sample material from a static test.
174 B. Wang, G. Lu / Thin-Walled Structures 40 (2002) 167–182

Fig. 8. Simulation of Tube 1 subjected to 300 m/s impact velocity. Frame times (µs): 0, 27.5, 55.0,
82.5, 110, 137.5, 165, 192.5, 220 and 247.5 (rebounded).

speeds. These patterns are largely verified by the tested samples in Figs. 2–6. The
initial wall thickness has a strong influence on the response. Fig. 9 shows a magnified
view of the sequence of deformation for Tube 1 with an impact velocity of 300 m/s.
The dynamic buckling is progressive starting from the striking end. Thickening of
the wall at the striking end can be clearly seen. The final thickness at the end appears
to have doubled compared with the undeformed thickness. Fig. 10 illustrates Tube
2 deforming under the same velocity. Mushrooming at the striking end is evident
as the wall end becomes thicker than the wall of the undeformed portion. The mush-
rooming effect also significantly alters the fold formations, as the first complete fold
does not occur at the very end. The increased wall thickness enhances the resistance
to buckling, thus shifting the first fold to a distance away from the tube end where
the thickening effect diminishes. This is clearly demonstrated by the tested sample
shown in Fig. 2. Though the stress wave plays an important role in the initial buck-
ling formation, it is the combined effect of the stress wave and mushrooming which
determines the final pattern. This is a clearly different phenomenon from that of
progressive buckling and plastic buckling reported previously, which always starts
from the loaded end. When the original undeformed wall thickness increases further,
mushrooming becomes predominant and the final deformation displays mushrooming
and wrinkling with no complete folds, as shown in Fig. 11 for Tube 3 at 300 m/s.
The influence of the striking velocity shows a similar trend with the deformation
mechanism evolving from folding at the striking end to mushrooming and folds at
a distance from the end, through to excessive mushrooming in the end region as the
velocity increases. For mushrooming, Figs. 12 and 13 illustrate the trend of the
B. Wang, G. Lu / Thin-Walled Structures 40 (2002) 167–182 175

Fig. 9. Magnified views of the striking end of Tube 1 under 300 m/s impact velocity. Frame times (µs):
0, 27.5, 55.0, 82.5, 110, 137.5, 165, 192.5, 220 and 247.5 (rebounded).
176 B. Wang, G. Lu / Thin-Walled Structures 40 (2002) 167–182

Fig. 10. Simulation on Tube 2 under 300 m/s impact velocity. Frame times (µs): 0, 27.5, 55.0, 82.5,
110, 137.5, 165, 192.5, and 220.
B. Wang, G. Lu / Thin-Walled Structures 40 (2002) 167–182 177

Fig. 11. Simulation on Tube 3 under 300 m/s impact velocity. Frame times (µs): 0, 37.5, 75,112.5, 150,
187.5, and 225.

change in tube length and wall thickness in terms of impact velocity for Tube 2.
Fig. 14 shows the deformed mesh in the vicinity of the striking end of Tube 3 at 100
µs after an impact at a striking velocity of 500 m/s. It gives an artificial illustration on
how the tube may deform. As revealed by the experiment, for all tubes impacted at
this high speed, there is inevitable fracture failures in the region of the striking end
where some of them split along the tube axis for several diameter lengths. This, of
course, casts doubts on the validity of the simulated results. This will be discussed
in Section 4.
The impact velocity effect is also illustrated in Fig. 15 on the buckling pattern,
showing the position of the first wrinkle in Tube 2 at various impact velocities.
178 B. Wang, G. Lu / Thin-Walled Structures 40 (2002) 167–182

Fig. 12. Percentage of reduction in the final length of Tube 1 under various striking velocities.

Fig. 13. Percentage of wall thickness increase at the first wrinkle in Tube 1 in terms of the striking
velocity.

Unlike the total length reduction shown in Fig. 12, which appears to be approxi-
mately linear in terms of striking velocity, this trend changes at the velocity of 400
m/s for wall thickening and the first wrinkle position in simulated results. Further
increase in the striking velocity seems to produce less effect, particularly regarding
the increase in wall thickness, indicating that the higher kinetic energy is mainly
dissipated by more fold formation, rather than mushrooming. Once again, one needs
to interpret this with great care. As fractures start to emerge at a velocity of around
200–300 m/s for mild steel tube samples used in this study, the final deforming
mechanism may differ for different materials.
Figs. 16 and 17 show the history of the simulated impact force at the striking
point and the energy dissipation in terms of the striking velocity for Tube 1. Interest-
ingly, the figures indicate that the duration of the impact event generally lasts for
0.2 ms irrespective of the striking velocity.
B. Wang, G. Lu / Thin-Walled Structures 40 (2002) 167–182 179

Fig. 14. Deformed mesh for the region close to the striking end of Tube 2 subjected to 500 m/s impact
velocity.

Fig. 15. Position of the first fold from the striking end in Tube 1 with respect to the striking velocity.

4. Discussion and conclusions

Both the experiments and the axisymmetric FE model confirm the effect of mush-
rooming in the axial dynamic crushing process of tubes with a relatively thick wall.
It shows that various modes of deformation will emerge and they can be significantly
different from those under a static or low speed loading condition. Generally, three
patterns of deformation may be expected: dynamic progressive folding for relatively
180 B. Wang, G. Lu / Thin-Walled Structures 40 (2002) 167–182

Fig. 16. History of the contact force for Tube 1 between the tube end and the rigid surface with respect
to the striking velocity.

thin tubes under a low impact speed; end mushrooming with folds formed at a dis-
tance from the striking end for all tubes at medium speeds, and mushrooming and
wrinkling for thick tubes at high speeds. For thick-walled tubes under high speed
axial impact, Taylor’s direct impact model [20] seems to be more appropriate, as
the dominant mode would be lateral deformation at the striking end. The difference
in the response between a thick-walled tube and a rod is probably the extensive
fractures and splitting associated with the former.
Direct correlation between the test results and simulations is difficult, mainly
because the FE code cannot model the dynamic fracture failures. The authors are
not aware of any well-verified dynamic fracture model in any commercial package.
The lack of such modelling capacity in commercial codes raises the question of their
validity when used for simulations in the higher velocity domain. However, for low
speed impact up to the range of 200–300 m/s where cracks are not commonly present,
the numerical analyses provide valid information and can help researchers obtain a
better understanding of the deformation process, which is difficult to achieve through
tests alone.
Furthermore, from a research perspective, if the ductility of the tube sample
material were increased by, for example, choosing a softer material, or by heating
the sample to an elevated temperature [21], it may be possible to reach even higher
speeds of impact without cracks forming. From this perspective, the present numeri-
cal simulation may cast some useful light on how the mechanisms form, though
B. Wang, G. Lu / Thin-Walled Structures 40 (2002) 167–182 181

Fig. 17. History of energy dissipation (kJ) in Tube 1 under various striking velocities.

material constants may need to be changed accordingly. Carrying out tests at elevated
temperatures remains a task for future research.

Acknowledgements

The authors wish to thank D. Ruan and M. Jewson for their help with the experi-
ments, and the Australian Research Council.

References

[1] Jones N. Structural impact. Cambridge: Cambridge Univesity Press, 1989.


[2] Andrews KRF, England GL, Ghani E. Classification of the axial collapse of cylindrical tubes under
quasi-static loading. Int J Mech Sci 1983;25:687.
[3] Alexander JM. An approximate analysis of the collapse of thin cylindrical columns. Q J Mech Appl
Math 1960;13:10.
[4] Abramowicz W, Jones N. Dynamic axial crushing of circular tubes. Int J Impact Engng 1984;2:263.
[5] Abramowicz W, Jones N. Dynamic progressive buckling of circular and square tubes. Int J Impact
Engng 1986;4:243.
[6] Grzebieta RH. An alternative method for determining the behaviour of round stocky tubes subjected
to axial crush loads. Thin-walled Structures 1990;9:66.
[7] Wierzbicki T, Bhat SU, Abramowicz W, Brodkin D. Alexander revisited—a two folding element
model of progressive crushing of tubes. Int J Solids Structures 1992;29:3269.
182 B. Wang, G. Lu / Thin-Walled Structures 40 (2002) 167–182

[8] Singace LA, Elsobky H, Reddy TY. On the eccentricity factor in the progressive crushing of tubes.
Int J Solids Structures 1995;32:3589.
[9] Guillow SR, Lu G, Grzebieta RH. Quasi-static axial compression of thin-walled circular aluminium
tubes. Int J Mech Sci in press.
[10] Karagiozova D, Jones N. Stress wave effects on the dynamic axial buckling of cylindrical shells
under impact. In: Jones N, editor. Structures under shock and impact V. Southampton: Computational
Mechanics Publications; 1998:201.
[11] Karagiozova D, Jones N. Dyanmic elastic-plastic buckling of circular cylindrical shells under axial
impact. Int J Solids Struct. 2000;37:2005.
[12] Karagiozova D, Alvies M, Jones N. Inertia effects in axisymmetrically deformed cylindrical shells
under axial impact. Int J Impact Engng 2000;24:1083.
[13] Wang R, Han M. Huang., Z. and Yan, Q., An experimental study on the dynamic axial plastic
buckling of cylindrical shell. Int J Impact Engng 1983;1:249.
[14] Chen C, Su X, Han M, Wang R. Elastic-plastic dynamics buckling and the stress wave in cylindrical
shells subjected to axial impact. In: Proc Int Symposium Intense Dynamic Loading and Its Effects,
Cengdu, 1992:543.
[15] Murase K, Jones N. The variation of modes in the dynamic axial plastic buckling of circular tubes.
In: Gupta NK, editor. Plasticity and impact mechanics. Oxford: Wiley Eastern; 1993:222.
[16] Florence AL, Goodier JN., Dynamic elastic-plastic buckling phenomena in a rod due to axial impact.
J Appl Mech (March);1968:80.
[17] Tvergaad V. On the transition from a diamond mode to an axisymmetric mode of collapse in cylindri-
cal shells. Int J Solids Structures 1983;19:845.
[18] Cowper GR, Symonds PS. Strain hardening and strain-rate effects in the impact loading of cantilever
beams. Brown University Division of Applied Mathematics Report No. 28, 1957.
[19] ABAQUS/Explicit manuals V5.8. Hibbitt, Karlsson & Sorensen, 1998.
[20] Taylor GI. The use of flat-ended projectile for determining dynamic yield stress I: theoretical con-
sideration. Proc R Soc 1948;A194:289.
[21] Hawkyard JB, Eaton D. And Johnson, W., The mean dynamic yield strength of copper and low
carbon steel at elevated temperatures from measurements of the mushrooming of flat-ended projec-
tiles. Int J Mech Sci 1969;11:313.

You might also like