Professional Documents
Culture Documents
International Journal of Crashworthiness
International Journal of Crashworthiness
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
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
© 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- behavior of the aluminum tubes was Material 24
displacement curves. (Mat_Piecewise-Linear-Plasticity), which is based on
piecewise linear isotropic plasticity and Von-Mises flow
MODELING DETAILS rule [8]. The engineering stress-strain curves of the tube
material were determined using four tensile specimens
This section gives the details of the modeling parameters machined from the walls of an extruded AA6063-T52
used for the simulation of quasi-static axial crushing tube. No significant difference was observed in the tensile
experiments conducted on a thin-walled square aluminum stress-strain curves of these four specimens. The true
tube [1]. The outer dimensions of the tube were 25.4 mm stress-true plastic strain curve of AA6063-T52 obtained
× 25.4 mm and the wall thickness was 1.6 mm (Fig. 1a). from the engineering stress-strain curve is shown in Fig. 2.
The tube was 140 mm long and contained a triggering Belytschko-Tsay (BT) shell elements with five
mechanism at its top end, which was also the loading end. integration points through the thickness and one
In most cases, the triggering mechanism was a chamfer integration point in the plane of the element were used in
on all four walls of the tube. The tube material was the analysis. In a few preliminary simulations, four different
aluminum alloy AA6063-T52. In the experiment, the tube mesh sizes, 2.0 mm × 2.0 mm, 2.5 mm × 2.5 mm, 3.5 mm
was held in a test fixture aligned with the loading axis, × 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
Downloaded by [Washington University in St Louis] at 11:57 09 April 2013
t = 1.6
140
25.4
25.4
(b)
(a)
Figure 1 (a) Geometric configuration of the square aluminum tube (dimensions are in mm) and (b) Finite element model.
250
200
100
50
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09
True plastic strain (mm/mm)
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 shows, there was very little difference between the force-
displacement responses. In the dynamic analysis reported
Explicit codes are designed for dynamic analysis, such as by Langseth et al. [2], the loading velocity was varied in
high-speed crash analysis, in which the event duration is the range of 5000–25000 mm/s. They found that the initial
very small and inertia plays a dominant role. Although peak force as well as the mean force increased with
explicit codes are commonly used for simulating quasi- increasing velocity. Furthermore, a change in the
static crush experiments, they require a large number of deformation mode occurred when the velocity was
time steps for quasi-static crush simulation. Two approaches increased from 20000 to 25000 mm/s. As shown in Fig. 3,
are recommended in the literature [9–10] for maintaining the force-displacement response at 10000 mm/s was very
quasi-static condition when an explicit code is used. In similar to that at 2000 mm/s. However, at 25000 mm/s,
the first approach, the mass density is scaled down to the maximum force was higher and the force-displacement
reduce the inertia forces, but the loading velocity is response was considerably different. Thus, at very high
increased to reduce the number of time steps. In the second loading velocities, it is possible that force-displacement
approach, the mass density is scaled up, but the loading response will change with increasing velocity and the
velocity is decreased, also to reduce the number of time analysis will not remain quasi-static.
steps. For aluminum alloys, the mass density is 2.7 ×10–6 Another consideration in the modeling for quasi-static
kg/mm3. Since scaling down the mass density of aluminum analysis is the mass of the loading plate. In the study
may result in very small time steps, the effect of scaling reported by Langseth et al. [2], no significant effect was
down the mass density was not considered in this study. observed when the mass of the loading plate was varied
Two approaches were used in this study. In one of these between 10 and 100 kg. A similar result was observed in
approaches, the mass density was not altered, but the this study in which the mass of the loading plate was
loading velocity was 1000, 1500, 2000, 10000 and 25000 either 10, 100 or 500 kg.
mm/s. In the second approach, the mass density was scaled Based on the various simulations conducted to validate
up to 2.7 × 10–3 kg/mm3 and the loading velocity was 25, quasi-static condition, the following modeling parameters
50, 75 and 100 mm/s. In these simulations, no physical were selected for the remainder of the study.
triggering mechanisms were employed to initiate the folding
process. Similar to the recommendations in Ref. 6, the • Scaled-up mass density = 2.7 × 10–3 kg/mm3
quasi-static condition was verified by performing the • Loading velocity = 100 mm/s
following checks: • Loading plate mass = 100 kg.
1. The ratio of total kinetic energy (KE) to the total internal
energy (IE) is less than 5 percent over the period of the CRUSH PARAMETERS
crushing process.
2. The crushing force-displacement response is The crush parameters determined from the force-
independent of the loading velocity. displacement responses are listed below.
Figure 3 shows the force-displacement responses of • Folding Initiation Force (FI): It is the force at which
several of the simulations performed to determine the the first fold formation is initiated. In some cases, it is
appropriate mass density-loading velocity combination also the maximum force observed in the force-
required to maintain quasi-static condition. As this figure displacement diagram.
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)
The quasi-static experiment conducted in Ref. 1 used Figure 4 (a) Chamfer on the outside wall of the tube
a machined 45° chamfer at the loading end of a AA6063- and (b) Numerical chamfering using elements of reduced
T52 aluminum tube. The dimensions of the tube in the thickness.
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. value (Table 1). The folding pattern for the 35.5° chamfer
Two different element heights were used: 2.5 mm and numerical model was asymmetric (Fig. 6), which was similar
1.93 mm, giving the chamfer angles as 35.5° and 45°, to that observed in the experiment. The folding pattern
respectively. Figure 5 compares the force-displacement of the 45° chamfer numerical model was symmetric, which
responses obtained from the numerical models and the was not observed in the experiment. Figure 6 also shows two
experiment. The 45° chamfer numerical model produced different stages of experimentally crushed tubes and the
a higher folding initiation force than the experimental folding patterns of 35.5° chamfer numerical model at the
value, but the folding initiation force in the 35.5° chamfer same stages of displacement. It can be seen that the number
numerical model was much closer to the experimental of lobes for each stage in the experiment and numerical
40000
35000
35.5 Degrees chamfer Experimental 45 Degrees chamfer
30000
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.
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
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
SA1.0: ta = 1.0 mm SA1.6: ta = 1.6 mm 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
SA1.2: ta = 1.2 mm SA1.8: ta = 1.8 mm 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
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 used to calculate the mean crush force. From the tension
equation developed by Abramowicz and Jones [14] was test data of the aluminum alloy used, the yield strength
used. They observed that this equation fitted equally well and the ultimate tensile strength were determined to be
to both symmetric and asymmetric folding patterns. 170 and 195 MPa, respectively. The strain-hardening
exponent was 0.089, which was calculated using a power
FM = 13.055 σo t5/3 b1/3 [3] law fit to the true stress-true plastic strain plot (Fig. 2).
In this equation, σo is the flow stress of the tube material. As shown in Fig. 10, the mean crush forces predicted
Two methods were employed here to calculate σo. In the by the Abramowicz-Jones equation are relatively close to
first method σY was used for σo, and in the second method, the numerically obtained values. In the symmetric folding
the following equation was used for σo [6]. area, the Abramowicz-Jones prediction using the second
method and numerical prediction are nearly the same. In
σY σ u the asymmetric folding area, the Abramowicz-Jones
σo = [4] prediction by either method is higher; however, it gives
n+1
the same trend as the numerical prediction. The
where, σY and σu are the yield strength and ultimate tensile Abramowicz-Jones equation based on Method 1, that is,
strength, respectively, and n is the strain-hardening σo = σY, is within ± 10 percent of the numerical results in
exponent of the tube material. Both these methods were both symmetric and asymmetric folding areas.
70000
Mahmood and Paluszny Stowell
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)
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)
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
(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 The others contained either a single triggering mechanism
end of the tube (Fig. 12), similar to the one reported in (35.5° chamfer, triangular hole pattern or geometric
Ref. 1. The hole diameter was 3.8 mm. In practice, these imperfection) or a combination of the triggering
holes can be either drilled or punched on all four walls of mechanisms. The cross sections of tubes in Models 1–8
the tube. The geometric imperfection considered here had sharp corners and the cross sections of tubes in Models
was a variation in mean dimensions of the tube along its 9–12 had rounded corners of 1.6 mm corner radius.
length. The mean dimensions of the perfect tube were For the sake of brevity, only the force-displacement
23.8 mm × 23.8 mm. The mean dimensions of the tube responses and the folding patterns of Models 1–4 are
with geometric imperfections were 24 mm × 24 mm for described in details. Figure 13 shows the force-displacement
the top 1/3 rd of its length, 23.6 mm × 23.6 mm for the responses obtained with these models and Fig. 14 shows
middle 1/3rd of its length, and 23.8 mm × 23.8 mm for a comparison of their folding patterns. The crush
the bottom 1/3rd of its length. In addition, the effect of performance values of all twelve models are listed in Table 3.
corner radius was also investigated. Twelve different
numerical models were created for the triggering Model 1: This model does not contain any physical
mechanism study (Table 3). Two of these models (Models triggering mechanism. The corners of the cross section
1 and 9) did not have any physical triggering mechanism. were also sharp. Even though there were no physical
45° chamfer
Downloaded by [Washington University in St Louis] at 11:57 09 April 2013
15° chamfer
Figure 11 Folding patterns of 1.6 mm thick tube triggered by 15, 25, 45 and 55° chamfer angle.
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
20000
15000
10000
5000
0
0 10 20 30 40 50 60 70 80 90
Displacement (mm)
1 and Model 2, Model 3 did not exhibit an initial high Model 4: The triggering mechanism in Model 4 was a
peak (Fig. 13). In Model 3, folding was initiated by the geometric imperfection. The folding initiation force of
collapse of the holes at the first folding initiation force, Model 4 was slightly lower than that of Model 1. Folding
which was observed at 19900 N. Thus, the folding initiation in Model 4 started at the geometric imperfection instead
force in this case was 31 percent lower than that of Model of at the loading end. After the folding was initiated, the
1 and 19 percent lower than that of Model 2. After folding force decreased gradually and the next force peak occurred
initiation, the average peak and valley forces of Model 3 at a much larger displacement than that in Model 1. The
were similar to their counterparts of Model 1. The folding tube wall deformed progressively forming alternating
pattern of Model 3 was symmetric (Fig. 14), which was inward and outward thick folds in two connecting edges,
similar to that of Model 1. much like in Model 2.
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
CONCLUSIONS
asymmetric. In some cases, a mixed folding pattern of International Journal of Crashworthiness 2004; 9: 73–87.
asymmetric and symmetric folds was also observed. 6. SANTOSA, S P, WIERZBICKI, T, HANSSEN, A G and LANGSETH,
The folding pattern as well as crush forces depended M. Experimental and numerical studies of foam-filled
on tube thickness. The folding pattern changed from sections. International Journal of Impact Engineering 2000;
24: 509–534.
symmetric to asymmetric as the tube thickness was
7. ALTENHOF, W, HARTE, A M and TURCHI, R. Experimental
increased; however, the transition from symmetric to
and numerical compressive testing of aluminum foam filled
asymmetric folding pattern also depended on the chamfer mild steel tubular hat sections. Proceedings 7th International
angle. For a given chamfer angle, the folding initiation LS-DYNA Users Conference, 2002.
force and the mean crush force increased with increasing 8. LS-DYNA Keyword User’s Manual, Version 960, Volume
tube thickness. The Abramowicz-Jones equation seems to II, Livermore Software Technology Corp., Livermore, CA,
predict the mean crush force of aluminum tubes reasonably 2001.
well for all tube thicknesses considered. 9. BATHE, K J, GUILLERMIN, O and CHEN, H Y. Advances in
nonlinear finite element analysis of automobiles. Computers
& Structures 1997; 64: 881–891.
REFERENCES
10. BATHE, K J, WALCZAK, J, GUILLERMIN, O, BOUZINOV, P A
1. BABBAGE, J and MALLICK, P K. Axial crush resistance of and CHEN, H Y. Advances in crush analysis, Computers &
aluminum composite hybrid tubes. Proceedings of 17th Structures 1999; 72: 31–47.
Downloaded by [Washington University in St Louis] at 11:57 09 April 2013
Annual Technical Conference of the American Society for 11. MATZENMILLER, A and SCHWEIZERHOF, K. Crashworthiness
Composites 2002, Paper No. 070. simulations of composite structures – a first step with
2. LANGSETH, M, HOPPERSTAD, O S and BERSTAD, T. explicit time integration. Nonlinear computational
Crashworthiness of aluminium extrusions: validation of mechanics-state of the art (Ed. P. Wriggers and W. Wagner),
numerical simulation, effect of mass ratio and impact Springer-Verlag, 1991.
velocity. International Journal of Impact Engineering 1999; 12. STOWELL, E Z. Compressive strength of flanges. NACA TN
22: 829–854. 2020. 1950.
3. KRAUSS, C A and LAANANEN D H. A parametric study of 13. MAHMOOD, H F and PALUSZNY, A. Design of thin walled
crush initiators for a thin-walled tube. International Journal columns for crash energy management-their strength and
of Vehicle Design 1994; 15: 385–401. mode of collapse. 1981. SAE Technical Paper No. 811302.
4. LEE, S, HAHN C, RHEE, M and OH, J. Effect of triggering 14. ABRAMOWICZ, W and JONES, N. Dynamic axial crushing of
on the energy absorption capacity of axially compressed square tubes. International Journal of Impact Engineering
aluminium tubes. Materials and Design 1999; 20: 31–40. 1984; 2: 179–208.
5. ARNOLD, B and ALTENHOF, W. Experimental observations on 15. LANGSETH, M and HOPPERSTAD, O S. Local buckling of
the crush characteristics of AA 6061 T4 and T6 structural square thin-walled aluminium extrusions. Thin-Walled
square tubes with and without circular discontinuities, Structures 1997; 27: 117–126.