This article was downloaded by: [Washington University in St Louis]

On: 09 April 2013, At: 11:57

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer
House, 37-41 Mortimer Street, London W1T 3JH, UK

International Journal of Crashworthiness

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/tcrs20

A numerical study on the quasi-static axial crush

characteristics of square aluminum tubes with
chamfering and other triggering mechanisms
H El-Hage , P K Mallick & N Zamani
a
University of Windsor, 401 Sunset Avenue, Windsor, Ontario, Canada N9B 3P4
b
University of Michigan-Dearborn, 4901 Evergreen Road, Dearborn, MI 48128, USA
c
University of Windsor, 401 Sunset Avenue, Windsor, Ontario, Canada N9B 3P4
Version of record first published: 08 Jul 2010.

To cite this article: H El-Hage , P K Mallick & N Zamani (2005): A numerical study on the quasi-static axial crush
characteristics of square aluminum tubes with chamfering and other triggering mechanisms, International Journal of
Crashworthiness, 10:2, 183-196

To link to this article: http://dx.doi.org/10.1533/ijcr.2005.0337

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to
anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss, actions,
claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or
indirectly in connection with or arising out of the use of this material.
 A numerical study on the quasi-static axial
crush characteristics of square aluminum
tubes with chamfering and other triggering
mechanisms
doi:10.1533/ijcr.2005.0337
1 2 1
H El-Hage , P K Mallick * and N Zamani
1
University of Windsor, 401 Sunset Avenue, Windsor, Ontario, Canada N9B 3P4
2
University of Michigan-Dearborn, 4901 Evergreen Road, Dearborn, MI 48128, USA
Downloaded by [Washington University in St Louis] at 11:57 09 April 2013

Abstract: This paper presents a numerical study on the effect of triggering mechanism on the load-
displacement characteristics and folding pattern of square aluminum tubes subjected to quasi-static
axial crushing. Chamfering was the principal triggering mechanism investigated. The other triggering
mechanisms included a triangular hole pattern, a geometric imperfection and combinations thereof.
The study has shown that triggering mechanism can significantly control the folding initiation force,
but its effect on mean crush force was relatively small. The effects of tube thickness and corner radius
were also considered. Both folding initiation force and mean crush force increase with tube thickness.
The load-displacement response does not depend appreciably on the corner radius.
Key words: Crush characteristics, triggering mechanisms, chamfering, aluminum tubes.
Chamfer /ˈʃæm.fɚ/ => chamfering (N) sự vát mép

INTRODUCTION aluminum tubes with a circular hole at their mid length.

In axial crushing of thin-walled tubes, a triggering
mechanism or a crush initiator is often used to initiate
progressive folding of the tube. Experimentally, the
triggering mechanism may be as simple as a chamfer
machined at the loading end of the tube or a series of
holes drilled on the walls close to the loading end of the
tube [1]. The triggering mechanism used in vehicle rails
is usually a series of convolutions or small cut outs located
strategically along their length. Langseth et al. [2] used a
geometric imperfection in the shape of a half cosine wave
on one of the walls of the tube as the triggering mechanism
in their numerical study. Krauss and Laananen [3]
experimentally examined the effectiveness of semi-circular
grooves, circular corner holes and diamond-shaped corner
holes as crush initiators. Lee et al. [4] did both experimental
and numerical studies on the crush characteristics of
rectangular aluminum tubes with side dents as crush
initiators. Arnold et al. [5] conducted both experimental
and numerical studies on the energy absorption of square
Corresponding Author:
P K Mallick
University of Michigan-Dearborn, 4901 Evergreen Road, Dearborn,
MI 48128, USA
Tel: +1 313 593 5119 Fax: +1 313 593 5386
Email: e-mail address: pkm@umich.edu
The primary purpose of this paper is to present a
numerical study examining the effects of chamfering and
other triggering mechanisms on quasi-static axial crush
performance of a thin-walled square aluminum tube. In
addition to the triggering mechanism study, the effect of
tube thickness upon the force-displacement response and
the folding mode was also investigated. The numerical
study was conducted using LS-DYNA, a commercial
explicit finite element code developed for simulating non-
linear events.
Several investigators in the past have used LS-DYNA
in crush simulation studies. For example, Langseth et al.
[2] conducted a numerical study using LS-DYNA and
validated the numerical results with the experimental data
obtained from static and dynamic tests on square aluminum
extrusions. Even though exact match between the predicted
and experimental curves was not achieved, they noted
that the maximum force and average force were predicted
with sufficient accuracy. On average, the ratios between
the predicted and the measured values were within ± 10
percent. Santosa et al. [6] used LS-DYNA to simulate
quasi-static crush tests of aluminium foam-filled square
aluminum extrusions. In this study, the predicted crushing
force was found to be in reasonable agreement with the
experimental results. Altenhof et al. [7] presented a
numerical study, also using LS-DYNA, on aluminum foam-
filled steel tubular hat sections. They also found reasonable

© Woodhead Publishing Ltd 0337 183 IJCrash 2005 Vol. 10 No. 2 pp. 183–195
 H El-Hage, P K Mallick and N Zamani
agreement between the experimental and simulated force-
displacement curves.
MODELING DETAILS
This section gives the details of the modeling parameters
used for the simulation of quasi-static axial crushing
experiments conducted on a thin-walled square aluminum
tube [1]. The outer dimensions of the tube were 25.4 mm
× 25.4 mm and the wall thickness was 1.6 mm (Fig. 1a).
The tube was 140 mm long and contained a triggering
mechanism at its top end, which was also the loading end.
In most cases, the triggering mechanism was a chamfer
on all four walls of the tube. The tube material was
aluminum alloy AA6063-T52. In the experiment, the tube
was held in a test fixture aligned with the loading axis,
Downloaded by [Washington University in St Louis] at 11:57 09 April 2013
and an axial load was applied using a thick steel loading
plate at the top end. The loading velocity in the experiment
was 12.7 mm/min.
For the boundary conditions, the bottom end of the
tube was assumed to be built-in and was constrained in
all degrees of freedom. The loading plate was modeled as
a rigid plate and its mass was assumed to be 100 kg (Fig.1b).
Except for simulations conducted to validate quasi-static
condition, the loading plate was displaced at a constant
loading velocity of 100 mm/sec in the negative Z-direction.
The Automatic-Nodes-To-Surfce algorithm [8] was used
at the interface between the loading plate and the loaded
end of the tube. Another contact algorithm, called
Automatic–Single-Surface [8], was used to prevent
interpenetration of folds during progressive folding of
the tube. The interface force between the loading plate
and the tube was used to obtain the force-displacement
responses.
The material model selected to represent the material
t = 1.6
140
25.4
25.4
(a)
Figure 1 (a) Geometric configuration of the square aluminum tube (dimensions are in mm) and (b) Finite element model.
IJCrash 2005 Vol. 10 No. 2
behavior of the aluminum tubes was Material 24
(Mat_Piecewise-Linear-Plasticity), which is based on
piecewise linear isotropic plasticity and Von-Mises flow
rule [8]. The engineering stress-strain curves of the tube
material were determined using four tensile specimens
machined from the walls of an extruded AA6063-T52
tube. No significant difference was observed in the tensile
stress-strain curves of these four specimens. The true
stress-true plastic strain curve of AA6063-T52 obtained
from the engineering stress-strain curve is shown in Fig. 2.
Belytschko-Tsay (BT) shell elements with five
integration points through the thickness and one
integration point in the plane of the element were used in
the analysis. In a few preliminary simulations, four different
mesh sizes, 2.0 mm × 2.0 mm, 2.5 mm × 2.5 mm, 3.5 mm
× 3.5 mm and 5.0 mm × 5.0 mm, were considered. The
folding patterns obtained with 2.0 mm × 2.0 mm and 2.5
mm × 2.5 mm mesh sizes were similar. Both produced
asymmetric folding pattern, which was also the folding
pattern observed in the experiment [1]. The folding patterns
obtained with 3.5 mm × 3.5 mm and 5.0 mm × 5.0 mm
mesh sizes were also asymmetric; however, the folds did
not form as smoothly as their fine mesh counterparts.
Since the 2.0 mm × 2.0 mm and 2.5 mm × 2.5 mm mesh
sizes produced smooth folds similar to those observed in
experiments [1], either of these two mesh sizes could be
used in simulation. The 2.5 mm × 2.5 mm mesh size was
selected in this study, since it would involve lower
computation time. It is worth noting that Langseth et al.
[2] found that 3.0 mm × 3.0 mm mesh size was sufficient
to capture the folding mechanism and force-displacement
response of square aluminum tubes. In their study, they
used the Belytschko-Tsay (BT) shell elements with nine
integration points through the thickness and one
integration point in the plane of the element.
100 kg rigid plate
Clamped support. Nodes are
constrained in all degrees of
freedom
(b)
184 doi:10.1533/ijcr.2005.0337 © Woodhead Publishing Ltd
 A numerical study on the quasi-static axial crush characteristics of square aluminum tubes
250
200
True stress (MPa) 150
100
50
0
0 0.01 0.02
Downloaded by [Washington University in St Louis] at 11:57 09 April 2013
Figure 2 True stress-true plastic strain diagram of the tube material (AA 6063-T52).
VALIDATION OF QUASI-STATIC CONDITION
Explicit codes are designed for dynamic analysis, such as
high-speed crash analysis, in which the event duration is
very small and inertia plays a dominant role. Although
explicit codes are commonly used for simulating quasi-
static crush experiments, they require a large number of
time steps for quasi-static crush simulation. Two approaches
are recommended in the literature [9–10] for maintaining
quasi-static condition when an explicit code is used. In
the first approach, the mass density is scaled down to
reduce the inertia forces, but the loading velocity is
increased to reduce the number of time steps. In the second
approach, the mass density is scaled up, but the loading
velocity is decreased, also to reduce the number of time
steps. For aluminum alloys, the mass density is 2.7 ×10–6
kg/mm3. Since scaling down the mass density of aluminum
may result in very small time steps, the effect of scaling
down the mass density was not considered in this study.
Two approaches were used in this study. In one of these
approaches, the mass density was not altered, but the
loading velocity was 1000, 1500, 2000, 10000 and 25000
mm/s. In the second approach, the mass density was scaled
up to 2.7 × 10–3 kg/mm3 and the loading velocity was 25,
50, 75 and 100 mm/s. In these simulations, no physical
triggering mechanisms were employed to initiate the folding
process. Similar to the recommendations in Ref. 6, the
quasi-static condition was verified by performing the
following checks:
1. The ratio of total kinetic energy (KE) to the total internal
energy (IE) is less than 5 percent over the period of the
crushing process.
2. The crushing force-displacement response is
independent of the loading velocity.
Figure 3 shows the force-displacement responses of
several of the simulations performed to determine the
appropriate mass density-loading velocity combination
required to maintain quasi-static condition. As this figure
© Woodhead Publishing Ltd doi:10.1533/ijcr.2005.0337
0.03 0.04 0.05 0.06 0.07 0.08 0.09
True plastic strain (mm/mm)
shows, there was very little difference between the force-
displacement responses. In the dynamic analysis reported
by Langseth et al. [2], the loading velocity was varied in
the range of 5000–25000 mm/s. They found that the initial
peak force as well as the mean force increased with
increasing velocity. Furthermore, a change in the
deformation mode occurred when the velocity was
increased from 20000 to 25000 mm/s. As shown in Fig. 3,
the force-displacement response at 10000 mm/s was very
similar to that at 2000 mm/s. However, at 25000 mm/s,
the maximum force was higher and the force-displacement
response was considerably different. Thus, at very high
loading velocities, it is possible that force-displacement
response will change with increasing velocity and the
analysis will not remain quasi-static.
Another consideration in the modeling for quasi-static
analysis is the mass of the loading plate. In the study
reported by Langseth et al. [2], no significant effect was
observed when the mass of the loading plate was varied
between 10 and 100 kg. A similar result was observed in
this study in which the mass of the loading plate was
either 10, 100 or 500 kg.
Based on the various simulations conducted to validate
quasi-static condition, the following modeling parameters
were selected for the remainder of the study.
• Scaled-up mass density = 2.7 × 10–3 kg/mm3
• Loading velocity = 100 mm/s
• Loading plate mass = 100 kg.
CRUSH PARAMETERS
The crush parameters determined from the force-
displacement responses are listed below.
• Folding Initiation Force (FI): It is the force at which
the first fold formation is initiated. In some cases, it is
also the maximum force observed in the force-
displacement diagram.
185 IJCrash 2005 Vol. 10 No. 2
 H El-Hage, P K Mallick and N Zamani

40000

Scaled-up density = 2.7 E-03 kg/mm^3 and loading velocity = 100 mm/s
35000
Original density = 2.7 E-06 kg/mm^3 and loading velocity = 2000 mm/s

30000 Original density = 2.7 E-06 kg/mm^3 and loading velocity = 10000 mm/s

Original density = 2.7 E-06 kg/mm^3 and loading velocity = 25000 mm/s
25000
Force (N)

20000

15000

10000
Downloaded by [Washington University in St Louis] at 11:57 09 April 2013

5000

0
0 10 20 30 40 50 60 70 80 90
Displacement (mm)

Figure 3 Force-displacement responses at different mass densities and loading velocities.

• Mean Crush Force:

Internal Energy @ 70 mm Crush Distance
FM =
70 mm α
Chamfered α
zone

MODELING FOR CHAMFERING

Chamfer is a beveled edge usually machined on the outside

surface of the tube. Since the end of the chamfer has a
sharp corner, it is difficult to model it unless a large number
of very small elements are used in the corner area. Following
the approach used by Matzenmiller and Schweizerhof
[11], chamfer zone in this study was modeled using
rectangular elements of progressively reduced thickness
as shown in Figure 4. The chamfer in the model is not on
the outside surface of the tube and the chamfer zone does
not have the shape of a right-angled triangle (Fig. 4a);
instead, it is on both inside and outside surfaces and has
t t
the shape of an isosceles triangle (Fig. 4b). The vertex
angle of this triangle represents the chamfer angle α. (a) (b)

The quasi-static experiment conducted in Ref. 1 used
a machined 45° chamfer at the loading end of a AA6063-
T52 aluminum tube. The dimensions of the tube in the
experiment are the same as shown in Fig. 1. The element
size in the numerical models was 2.5 mm × 2.5 mm. The
chamfer zone was represented by a 0.8 mm thick elements.
Two different element heights were used: 2.5 mm and
1.93 mm, giving the chamfer angles as 35.5° and 45°,
respectively. Figure 5 compares the force-displacement
responses obtained from the numerical models and the
experiment. The 45° chamfer numerical model produced
a higher folding initiation force than the experimental
value, but the folding initiation force in the 35.5° chamfer
numerical model was much closer to the experimental
Figure 4 (a) Chamfer on the outside wall of the tube
and (b) Numerical chamfering using elements of reduced
thickness.
value (Table 1). The folding pattern for the 35.5° chamfer
numerical model was asymmetric (Fig. 6), which was similar
to that observed in the experiment. The folding pattern
of the 45° chamfer numerical model was symmetric, which
was not observed in the experiment. Figure 6 also shows two
different stages of experimentally crushed tubes and the
folding patterns of 35.5° chamfer numerical model at the
same stages of displacement. It can be seen that the number
of lobes for each stage in the experiment and numerical

IJCrash 2005 Vol. 10 No. 2 186 doi:10.1533/ijcr.2005.0337 © Woodhead Publishing Ltd

 A numerical study on the quasi-static axial crush characteristics of square aluminum tubes

40000

35000
35.5 Degrees chamfer Experimental 45 Degrees chamfer

30000

Force (N) 25000

20000

15000

10000

5000

0
0 10 20 30 40 50 60 70 80
Displacement (mm)
Downloaded by [Washington University in St Louis] at 11:57 09 April 2013

Figure 5 Comparison of force-displacement responses of 35.5° and 45° chamfer numerical models with experimental
force-displacement response.

Figure 6 Comparison of experimental and numerical folding patterns.

model was closely matched. In order to examine if a smaller EFFECT OF THICKNESS

element size would further improve the correlation with
the experiment, a new 35.5° chamfer numerical model
was constructed with 2.0 mm × 2.0 mm element size. A
small improvement in the folding initiation force (Table
1) was observed; however, the force-displacement response
and the folding pattern did not change appreciably.
The wall thickness (t) of the aluminum tube was varied
between 1.0 mm to 2 mm and the tube length was 127
mm, giving a range of width-to-thickness ratios from 12.7
to 25.4 and length-to-thickness ratios ranging from 63.5
to 127. The chamfer angle in these tubes was 35.5°. All of
the tubes considered in this study exhibited progressive

© Woodhead Publishing Ltd doi:10.1533/ijcr.2005.0337 187 IJCrash 2005 Vol. 10 No. 2

 H El-Hage, P K Mallick and N Zamani

Table 1 Experimental vs. numerical results 1.8 and 2 mm thick tubes folded with an asymmetric
pattern. Thus, there was a change in folding pattern from
Description Element size Folding Displacement symmetric to asymmetric between 1.4 and 1.5 mm.
(mm × mm) initiation at folding However, as shown in Fig. 8, the folding initiation force
force (N) initiation as well as the mean crush force increased with increasing
(mm)
thickness. Furthermore, the maximum displacement or
Numerical Model- 2.5 × 2.5 27,300 4.23 crush stroke was different for symmetric and asymmetric
45° chamfer fold patterns, and in each category, it decreased with
Numerical Model- 2.5 × 2.5 24,600 3.43 increasing thickness due to increasing lobe thickness. The
35.5° chamfer crush stroke of the 2.0 mm thick tube was only 35 mm,
Numerical Model- 2×2 25,800 4.19 which is less than half of the crush stroke of the 1.5 mm
35.5° chamfer thick tube.
Experiment – 25,130 4.54
The folding initiation force and the mean crush force
are plotted as a function of tube thickness in Fig. 9 and
folding; however, as Fig. 7 shows, 1.0, 1.2 and 1.4 mm Fig. 10, respectively. The transition from symmetric to
thick tubes folded with a symmetric pattern, while 1.6, asymmetric folding pattern occurred at a thickness between
Downloaded by [Washington University in St Louis] at 11:57 09 April 2013

SA1.0: ta = 1.0 mm
SA1.2: ta = 1.2 mm
SA1.6: ta = 1.6 mm
SA1.8: ta = 1.8 mm
1.4 and 1.5 mm. For the prediction of the folding initiation
force, two equations were considered, namely the Stowell
equation [12] and the Mahmood-Paluszny equation [13].
The Abramowicz-Jones equation [14] was used for the
prediction of mean crush force.
The Stowell equation has been used previously by
Langseth and Hopperstad [15] to predict the ultimate
load of thin-walled aluminum extrusions. Assuming that
each side of the tube can be represented as a rectangular
plate with simply supported edges, the Stowell equation
is given by
4π 2 Es  t3   1 + 3 Es 
FI =   2 + 2  [1]
9  
b  4 4 Et 
where, FI = Folding initiation force
E s and E t = secant modulus and tangent modulus,
respectively
t = tube thickness
b = width
The secant modulus and tangent modulus for the aluminum
alloy used in the tube were calculated as 8500 MPa and
828 MPa, respectively. These values were determined at
σ = 1 (σ Y + σ u ) , where σY and σu are yield strength
2
and tensile strength of the tube material, respectively.
The Mahmood-Paluszny equation [13] for aluminum
tubes can be written as

FI = 568 t1.86 b0.14 σ Y0.57 [2]

SA1.4: ta = 1.4 mm SA2.0: ta = 2.0 mm
Figure 7 Symmetric folding pattern of 1.0, 1.2 and
1.4 mm thick tubes (left) and asymmetric folding patterns
of 1.6, 1.8 and 2 mm thick tubes (right) (Triggering
mechanism: 35.5° numerical chamfer)
As shown in Fig. 9, the Mahmood-Paluszny equation
predicts folding initiation force close to the numerical
prediction at low thicknesses, but much higher folding
initiation force than the numerical prediction at high
thicknesses. The Stowell equation, on the other hand,
gives much lower folding initiation force than the numerical
prediction at low thicknesses, but the difference between
the Stowell equation and numerical prediction becomes
smaller as the thickness increases. At 2 mm tube thickness,
the Stowell equation and the numerical prediction are
nearly equal.

IJCrash 2005 Vol. 10 No. 2 188 doi:10.1533/ijcr.2005.0337 © Woodhead Publishing Ltd

 A numerical study on the quasi-static axial crush characteristics of square aluminum tubes

30000
SA1.0: Thickness = 1.0 mm, 35.5 degrees chamfer
SA1.2: Thickness = 1.2 mm, 35.5 degrees chamfer
25000 SA1.4 Thickness = 1.4 mm, 35.5 degrees chamfer

20000
Force (N)

15000

10000

5000

0
0 10 20 30 40 50 60 70 80 90
Downloaded by [Washington University in St Louis] at 11:57 09 April 2013

Displacement (mm)
(a)

40000
SA1.6: Thickness = 1.6 mm, 35.5 degrees chamfer
35000 SA1.8: Thickness = 1.8 mm, 35.5 degrees chamfer
SA2.0: Thickness = 2.0 mm, 35.5 degrees chamfer
30000

25000
Force (N)

20000

15000

10000

5000

0
0 10 20 30 40 50 60 70 80 90
Displacement (mm)
(b)

Figure 8 Force-displacement responses of (a) 1.0, 1.2 and 1.4 mm thick tubes (symmetric folding pattern) and (b) 1.6, 1.8
and 2 mm thick tubes (asymmetric folding pattern).

For the prediction of mean crush force, the following
equation developed by Abramowicz and Jones [14] was
used. They observed that this equation fitted equally well
to both symmetric and asymmetric folding patterns.
FM = 13.055 σo t5/3 b1/3
In this equation, σo is the flow stress of the tube material.
Two methods were employed here to calculate σo. In the
first method σY was used for σo, and in the second method,
the following equation was used for σo [6].
σY σ u
σo =
n+1
where, σY and σu are the yield strength and ultimate tensile
strength, respectively, and n is the strain-hardening
exponent of the tube material. Both these methods were
used to calculate the mean crush force. From the tension
test data of the aluminum alloy used, the yield strength
and the ultimate tensile strength were determined to be
170 and 195 MPa, respectively. The strain-hardening
exponent was 0.089, which was calculated using a power
[3] law fit to the true stress-true plastic strain plot (Fig. 2).
As shown in Fig. 10, the mean crush forces predicted
by the Abramowicz-Jones equation are relatively close to
the numerically obtained values. In the symmetric folding
area, the Abramowicz-Jones prediction using the second
method and numerical prediction are nearly the same. In
the asymmetric folding area, the Abramowicz-Jones
[4] prediction by either method is higher; however, it gives
the same trend as the numerical prediction. The
Abramowicz-Jones equation based on Method 1, that is,
σo = σY, is within ± 10 percent of the numerical results in
both symmetric and asymmetric folding areas.

© Woodhead Publishing Ltd doi:10.1533/ijcr.2005.0337 189 IJCrash 2005 Vol. 10 No. 2

 H El-Hage, P K Mallick and N Zamani

70000
Mahmood and Paluszny Stowell

60000 Numerical prediction

Folding initiation force (N)

50000
Symmetric

40000

30000

20000

10000 Asymmetric

0
1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2
Tube thickness (mm)

Figure 9 Variation of folding initiation force with tube thickness.

Downloaded by [Washington University in St Louis] at 11:57 09 April 2013

25000
Abramowicz and Jones (Method 1)
Abramowicz and Jones (Method 2)
Numerical prediction
20000
Mean crush force (N)

15000

10000

5000

Symmetric Asymmetric

0
1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.1
Tube thickness (mm)

Figure 10 Variation of mean crush force with tube thickness.

EFFECT OF CHAMFER ANGLE between 1.4 and 1.5 mm tube thickness. For 45 and 55°
chamfer angles, transition from symmetric to asymmetric
To examine the effect of chamfer angle on the folding folding occurred between 1.6 and 1.8 mm.
pattern, it was decided to conduct simulations on 1, 1.2. Table 2 lists the folding initiation forces and mean crush
1.4, 1.5, 1.6 1.8 and 2 mm thick tubes, each with five forces at various chamfer angles. For each tube thickness,
different chamfer angles, namely 15, 25, 35.5, 45 and 55°. the folding initiation force increased with increasing
The chamfer zone height and the number of elements chamfer angle; however, the mean crush force and energy
used in the chamfer zone for various chamfer angles are absorbed did not show any particular trend. For all practical
listed in Table 2. For long chamfer zones, three elements purposes, it can be assumed that the mean crush force
were used, whereas for short chamfer zones, only one and energy absorption of a tube of given thickness was
element was sufficient to model the chamfer zone. not affected significantly by chamfer angle. However, as
For 1.8 and 2 mm thick tubes, the folding pattern was Table 2 shows, for each chamfer angle, both folding
asymmetric at all chamfer angles. Except for one case, the initiation force as well as mean crush force increased with
folding pattern for 1, 1.2 and 1.4 mm thick tubes were increasing tube thickness.
symmetric at all chamfer angles. For 1.5 and 1.6 mm tubes,
the folding pattern was asymmetric up to 35.5° chamfer
angle and changed to symmetric mode above 45° chamfer OTHER TRIGGERING MECHANISMS
angle (Table 2). Figure 11 shows the change from
asymmetric to symmetric folding pattern for 1.6 mm thick In this part of the numerical study, the effect of two other
tubes. For 15, 25 and 35.5° chamfer angles, transition triggering mechanisms, namely a triangular pattern of holes
from symmetric folding to asymmetric folding occurred and a geometric imperfection has been investigated. The

IJCrash 2005 Vol. 10 No. 2 190 doi:10.1533/ijcr.2005.0337 © Woodhead Publishing Ltd

 A numerical study on the quasi-static axial crush characteristics of square aluminum tubes

Table 2 Crush characteristics of aluminum tubes with various chamfer angles

Tube Chamfer Chamfer Number Folding Mean Folding

thickness angle height of initiation crush pattern(1)
(mm) (degrees) (mm) elements force (N) force(2) (N)
1.0 15 3.8 2 12,240 6,957 S
25 2.25 1 13,321 6,871 S
35.5 1.56 1 14,200 7,067 S
45 1.2 1 14,332 7,114 S
55 0.96 1 15,875 7,714 S
1.2 15 4.56 2 18,583 7,829 A
25 2.7 1 17,157 9,357 S
35.5 1.87 1 18,100 9,511 S
45 1.45 1 18,945 9,529 S
55 1.15 1 19,532 9,471 S
1.4 15 5.32 3 19,888 11,671 S
25 3.15 2 24,071 12,386 S
Downloaded by [Washington University in St Louis] at 11:57 09 April 2013

35.5 2.19 1 22,500 12,248 S

45 1.69 1 23,090 12,029 S
55 1.34 1 23,738 12,557 S
1.5 15 5.7 3 22,901 11,114 A
25 3.38 2 24,545 11,428 A
35.5 2.34 1 24,200 11,773 A
45 1.8 1 25,232 13,986 S
55 1.44 1 26,074 14,542 S
1.6 15 6.07 3 21,040 12,286 A
25 3.61 2 25,664 12,357 A
35.5 2.5 1 24,600 13,234 A
45 1.93 1 27,566 15,842 S
55 1.54 1 28,394 16,657 S
1.8 15 6.84 3 26,720 16,900 A
25 4.05 2 29,563 17,157 A
35.5 2.81 1 28,000 16,529 A
45 2.17 1 31,037 16,114 A
55 1.73 1 33,673 19,085 A
2.0 15 7.59 3 28,334 17,986 A
25 4.51 2 32,744 20,685 A
35.5 3.12 2 34,400 20,006 A
45 2.41 1 34,139 19,171 A
55 1.92 1 34,908 18,843 A

(1) S = Symmetric, A = Asymmetric, (2) Based on internal energy @ 70 mm, except for 1.8 and 2 mm thickness. For 1.8 and 2 mm
thickness, based on internal energy @ 40 and 35 mm, respectively.

triangular pattern of holes was located near the loaded
end of the tube (Fig. 12), similar to the one reported in
Ref. 1. The hole diameter was 3.8 mm. In practice, these
holes can be either drilled or punched on all four walls of
the tube. The geometric imperfection considered here
was a variation in mean dimensions of the tube along its
length. The mean dimensions of the perfect tube were
23.8 mm × 23.8 mm. The mean dimensions of the tube
with geometric imperfections were 24 mm × 24 mm for
the top 1/3 rd of its length, 23.6 mm × 23.6 mm for the
middle 1/3rd of its length, and 23.8 mm × 23.8 mm for
the bottom 1/3rd of its length. In addition, the effect of
corner radius was also investigated. Twelve different
numerical models were created for the triggering
mechanism study (Table 3). Two of these models (Models
1 and 9) did not have any physical triggering mechanism.
The others contained either a single triggering mechanism
(35.5° chamfer, triangular hole pattern or geometric
imperfection) or a combination of the triggering
mechanisms. The cross sections of tubes in Models 1–8
had sharp corners and the cross sections of tubes in Models
9–12 had rounded corners of 1.6 mm corner radius.
For the sake of brevity, only the force-displacement
responses and the folding patterns of Models 1–4 are
described in details. Figure 13 shows the force-displacement
responses obtained with these models and Fig. 14 shows
a comparison of their folding patterns. The crush
performance values of all twelve models are listed in Table 3.
Model 1: This model does not contain any physical
triggering mechanism. The corners of the cross section
were also sharp. Even though there were no physical

© Woodhead Publishing Ltd doi:10.1533/ijcr.2005.0337 191 IJCrash 2005 Vol. 10 No. 2

 H El-Hage, P K Mallick and N Zamani
Downloaded by [Washington University in St Louis] at 11:57 09 April 2013
15° chamfer
25° chamfer
Figure 11 Folding patterns of 1.6 mm thick tube triggered by 15, 25, 45 and 55° chamfer angle.
8.0 mm 8.0 mm
3.7 mm
5.0 mm
Figure 12 Triangular hole pattern used as one of the
triggering mechanisms (hole diameter = 3.8 mm).
triggering mechanisms in this model, the tube walls folded
progressively, forming continuous inward and outward
thin folds in a symmetric pattern. The folding initiation
force was 28800 N, at which folding was initiated in the
tube. The triggering in this case may have occurred due
to lack of numerical precision of the software, hardware
or both.
IJCrash 2005 Vol. 10 No. 2
45° chamfer
55° chamfer
Model 2: This model contained a 35.5° chamfered end.
The folding initiation force, in this case, was 24600 N,
which was significantly lower than that in Model 1. The
mean crush force was also lower with the chamfered tube.
Before the folding initiation force was reached, there was
a smaller peak at which the 0.8 thick mm elements in the
chamfered zone started to collapse. After the folding
initiation force was reached, the tube walls deformed
progressively forming an asymmetric folding pattern of
alternating inward and outward thick folds in two
connecting edges. The relatively long folding length or
the relatively long distance between two consecutive
hinges decreased the mean crush force. The force range
in the progressive folding zone was also decreased. It is
interesting to note that the folding pattern and the lobe
shapes of Model 2 were different from those observed in
Model 1.
Model 3: In Model 3, a triangular pattern of holes was
used as the physical triggering mechanism. Unlike Model
192 doi:10.1533/ijcr.2005.0337 © Woodhead Publishing Ltd
 A numerical study on the quasi-static axial crush characteristics of square aluminum tubes

Table 3 Effect of triggering mechanism on folding

Model Corner Triggering Folding initiation Displacement at Mean crush Folding

no. radius mechanism force, FI folding initiation force, FM pattern
(mm) (N) (mm) (N)
1 0 None 28,800 3.67 16,050 Symmetric
2 0 35.5° Chamfer 24,600 3.93 12,675 Asymmetric
3 0 Hole pattern 19,900 0.953 15,570 Symmetric
4 0 Geometric 27,700 2.28 14,230 Asymmetric
imperfection
5 0 35.5° Chamfer + 22,900 3.38 13,270 Asymmetric
geometric
imperfection
6 0 35.5° chamfer + 17,500 4.03 16,070 Symmetric
hole pattern
7 0 Hole pattern + 19,700 0.986 15,780 Symmetric
geometric
Downloaded by [Washington University in St Louis] at 11:57 09 April 2013

imperfection
8 0 35.5° chamfer + 14,800 1.08 15,790 Symmetric
hole pattern +
geometric
imperfection
9 1.6 None 27,000 2.87 16,915 Mixed
10 1.6 35.5° Chamfer 22,400 3.82 12,075 Asymmetric
11 1.6 Hole pattern 17,900 0.981 13,905 Mixed
12 1.6 35.5° Chamfer + 16,500 3.88 13,490 Mixed
hole pattern

40000
Model 1: No physical trigger
35000
Model 2: 35.5 degrees chamfer

Model 3: Hole pattren

30000
Model 4: Geometric imperfection
25000
Force (N)

20000

15000

10000

5000

0
0 10 20 30 40 50 60 70 80 90
Displacement (mm)

Figure 13 Force-displacement diagrams of numerical models 1, 2, 3 and 4.

1 and Model 2, Model 3 did not exhibit an initial high
peak (Fig. 13). In Model 3, folding was initiated by the
collapse of the holes at the first folding initiation force,
which was observed at 19900 N. Thus, the folding initiation
force in this case was 31 percent lower than that of Model
1 and 19 percent lower than that of Model 2. After folding
initiation, the average peak and valley forces of Model 3
were similar to their counterparts of Model 1. The folding
pattern of Model 3 was symmetric (Fig. 14), which was
similar to that of Model 1.
Model 4: The triggering mechanism in Model 4 was a
geometric imperfection. The folding initiation force of
Model 4 was slightly lower than that of Model 1. Folding
in Model 4 started at the geometric imperfection instead
of at the loading end. After the folding was initiated, the
force decreased gradually and the next force peak occurred
at a much larger displacement than that in Model 1. The
tube wall deformed progressively forming alternating
inward and outward thick folds in two connecting edges,
much like in Model 2.

© Woodhead Publishing Ltd doi:10.1533/ijcr.2005.0337 193 IJCrash 2005 Vol. 10 No. 2

 H El-Hage, P K Mallick and N Zamani

ELE2C: Aluminum tube [SA2] Model 2: Square aluminum tube [SA2]

Time = 1.016 Time = 1.016 folding propagation were close to the folding initiation
force. Thus, from the standpoint of crush management,
the triangular hole pattern can be considered a better
triggering mechanism than chamfering.
A comparison of the load-displacement responses of
models with sharp corners and rounded corners showed
that the general behavior of the load-displacement response
did not change appreciably. However, Models 11 and 12
started to fold with symmetric mode of folding as in the
case of Models 1, 3 and 8, but, after one or two symmetric
fold formation, changed to asymmetric mode. Thus,
Model 1 Model 2
Models 9, 11 and 12 showed a mixed mode of folding
Model 3: Square aluminum tube [SA2]
Time = 1.100
Model 4: Square aluminum tube [SA2]
(Fig. 15). The folding initiation forces for the tubes with
Time = 1.06
rounded corners were also lower than the folding initiation
forces for the tubes with sharp corners.
Downloaded by [Washington University in St Louis] at 11:57 09 April 2013

Model 9: Square Aluminum tube [SA2] Model 10: Square Aluminum tube [SA2]
Time 1.112 Time 1.008
Model 9 Model 10

Model 3 Model 4
Model 5: Square aluminum tube [SA2]
Time = 1.008
Model 6: Square aluminum tube [SA2]
Time = 1.108

Model 11: Square Aluminum tube [SA2] Model 12: Square Aluminum tube [SA2]
Time 1.092 Time 1.060

Model 5 Model 6

Model 7: Square aluminum tube [SA2] Model 8: Square aluminum tube [SA2]
Time = 1.016 Time = 1.008

Model 12
Model 11

Figure 15 Folding patterns of Models 9–12 (tubes with

rounded corners).

CONCLUSIONS

Progressive folding in square aluminum tubes can be

Model 7 Model 8
Figure 14 Folding patterns of Models 1–8 (tubes with
sharp corners).
Table 3 gives the folding initiation forces, mean crush
forces and folding patterns of all twelve models. The folding
initiation force and the displacement at folding initiation
were very much dependent on the triggering mechanism.
The presence of physical triggering mechanism helped
reduce the folding initiation force. The lowest folding
initiation force was obtained with the triangular hole
pattern. In this case, the forces required for progressive
triggered using a variety of crush initiators. Physical crush
initiators such as chamfers and hole patterns are often
used in experiments. This study has shown that minor
geometric imperfection in the tube can also initiate
progressive folding. This study has also shown that
progressive folding in numerical models does not require
the presence of a physical triggering mechanism, since
lack of numerical precision can also initiate progressive
folding.
Both the force-displacement response and the folding
pattern are influenced by the triggering mechanism.
Depending on the triggering mechanism, two principal
types of folding patterns were observed: symmetric and

IJCrash 2005 Vol. 10 No. 2 194 doi:10.1533/ijcr.2005.0337 © Woodhead Publishing Ltd

 A numerical study on the quasi-static axial crush characteristics of square aluminum tubes
asymmetric. In some cases, a mixed folding pattern of
asymmetric and symmetric folds was also observed.
The folding pattern as well as crush forces depended
on tube thickness. The folding pattern changed from
symmetric to asymmetric as the tube thickness was
increased; however, the transition from symmetric to
asymmetric folding pattern also depended on the chamfer
angle. For a given chamfer angle, the folding initiation
force and the mean crush force increased with increasing
tube thickness. The Abramowicz-Jones equation seems to
predict the mean crush force of aluminum tubes reasonably
well for all tube thicknesses considered.
REFERENCES
1. BABBAGE, J and MALLICK, P K. Axial crush resistance of
aluminum composite hybrid tubes. Proceedings of 17th
Downloaded by [Washington University in St Louis] at 11:57 09 April 2013
Annual Technical Conference of the American Society for
Composites 2002, Paper No. 070.
2. LANGSETH, M, HOPPERSTAD, O S and BERSTAD, T.
Crashworthiness of aluminium extrusions: validation of
numerical simulation, effect of mass ratio and impact
velocity. International Journal of Impact Engineering 1999;
22: 829–854.
3. KRAUSS, C A and LAANANEN D H. A parametric study of
crush initiators for a thin-walled tube. International Journal
of Vehicle Design 1994; 15: 385–401.
4. LEE, S, HAHN C, RHEE, M and OH, J. Effect of triggering
on the energy absorption capacity of axially compressed
aluminium tubes. Materials and Design 1999; 20: 31–40.
5. ARNOLD, B and ALTENHOF, W. Experimental observations on
the crush characteristics of AA 6061 T4 and T6 structural
square tubes with and without circular discontinuities,
© Woodhead Publishing Ltd doi:10.1533/ijcr.2005.0337
International Journal of Crashworthiness 2004; 9: 73–87.
6. SANTOSA, S P, WIERZBICKI, T, HANSSEN, A G and LANGSETH,
M. Experimental and numerical studies of foam-filled
sections. International Journal of Impact Engineering 2000;
24: 509–534.
7. ALTENHOF, W, HARTE, A M and TURCHI, R. Experimental
and numerical compressive testing of aluminum foam filled
mild steel tubular hat sections. Proceedings 7th International
LS-DYNA Users Conference, 2002.
8. LS-DYNA Keyword User’s Manual, Version 960, Volume
II, Livermore Software Technology Corp., Livermore, CA,
2001.
9. BATHE, K J, GUILLERMIN, O and CHEN, H Y. Advances in
nonlinear finite element analysis of automobiles. Computers
& Structures 1997; 64: 881–891.
10. BATHE, K J, WALCZAK, J, GUILLERMIN, O, BOUZINOV, P A
and CHEN, H Y. Advances in crush analysis, Computers &
Structures 1999; 72: 31–47.
11. MATZENMILLER, A and SCHWEIZERHOF, K. Crashworthiness
simulations of composite structures – a first step with
explicit time integration. Nonlinear computational
mechanics-state of the art (Ed. P. Wriggers and W. Wagner),
Springer-Verlag, 1991.
12. STOWELL, E Z. Compressive strength of flanges. NACA TN
2020. 1950.
13. MAHMOOD, H F and PALUSZNY, A. Design of thin walled
columns for crash energy management-their strength and
mode of collapse. 1981. SAE Technical Paper No. 811302.
14. ABRAMOWICZ, W and JONES, N. Dynamic axial crushing of
square tubes. International Journal of Impact Engineering
1984; 2: 179–208.
15. LANGSETH, M and HOPPERSTAD, O S. Local buckling of
square thin-walled aluminium extrusions. Thin-Walled
Structures 1997; 27: 117–126.
195 IJCrash 2005 Vol. 10 No. 2
 Downloaded by [Washington University in St Louis] at 11:57 09 April 2013

IJCrash 2005 Vol. 10 No. 2

196
© Woodhead Publishing Ltd