Vous êtes sur la page 1sur 7

ARTICLE IN PRESS

Thin-Walled Structures 44 (2006) 701707 www.elsevier.com/locate/tws

Effect of transverse impact on buckling behavior of compressed column


Azhari Sastranegara1, Tadaharu Adachi, Akihiko Yamaji
Department of Mechanical Sciences and Engineering, Tokyo Institute of Technology, 2-12-1, O-okayama, Meguro-ku, Tokyo 152-8552, Japan Received 25 August 2005; accepted 11 April 2006 Available online 21 June 2006

Abstract The effect of transverse impact on the buckling behavior of a column under axial compressive load was numerically investigated with nite-element analysis. An axially compressed column was subjected to transverse-impact that instantaneously reduced its structural stiffness. Due to this impact, the column lost its stability and buckled even when the axial load was lower than Eulers buckling load. A transverse-impact load like a half-sine wave with different peaks and durations was applied to the column. We found that the critical condition to generate buckling was determined by the external work of the transverse-impact load. The velocity of column deformation increased as the maximum load for transverse-impact increased. Therefore, deformation during the post-buckling process might be controlled by transverse-impact. r 2006 Elsevier Ltd. All rights reserved.
Keywords: Column buckling; Transverse impact; Buckling load; Post-buckling; Finite element analysis

1. Introduction Safety in a collision is one of the main features in designing modern vehicles. Vehicles structures are designed to collapse, so that they can absorb collision energy [13]. In practice, imperfections are generally introduced to the structure to enable progressive buckling that will increase absorbed energy. Well-known examples of imperfections are dents or bellows in the side members. Fundamentally, these imperfections reduce the local stiffness so that a vehicle can collapse or buckle easily when impact occurs. However a very stiff vehicle structure is a primary design consideration required to improve driving performance, and lower vibration and noise. Therefore, these stiffness requirements for vehicle structures contradict each other. In practical design, stiffness must be compromised to satisfy both requirements. Since the demand for higher speeds and superior driving performance is expected to continually increase in the future, a stiffer vehicle structure will become very
Corresponding author. Tel.: +81 3 5734 2156; fax: +81 3 5734 2893.

important. Therefore, we have suggested an alternative way of maintaining a high degree of stiffness while improving energy absorption. The concept of this research is to improve energy absorption without reducing

E-mail address: adachi@mech.titech.ac.jp (T. Adachi). Present address: Corporate Research and Development Center, NSK Ltd., 1-5-50, Kugenuma Shinmei, Fujisawa-Shi Kanagawa 251-8501, Japan.
1

Fig. 1. Analytical model

0263-8231/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.tws.2006.04.010

ARTICLE IN PRESS
702 A. Sastranegara et al. / Thin-Walled Structures 44 (2006) 701707

structural stiffness or strength beforehand by providing external disturbance only when impact occurs [46]. To improve the absorption of structural energy, it is necessary
Table 1 Parameters in analysis Parameter Qmax (N) T (ms) xQ, (mm) P (N) Value 1002000 1, 0.5, 0.2, 10.1, 0.05 80, 160, 240, 320, 360 275, 300, 315

to understand how buckling can be controlled in the behavior of a compressed structure. Column buckling caused solely by static axial load [7] and beam bending caused solely by transverse-impact [8,9] has already been discussed in much of the literature. It is well known that buckling can occur under a load lower than Eulers buckling load when an external disturbance or pre-stress is applied. Several researchers have also studied the transverse-impact problems that occur simultaneously with static axial load [1013]. However, effect of transverse-impact on column buckling under axial load that has focused on energy absorption has not been investigated yet.
10 Axial displacement u [mm]
Qmax: 500 N

10 Axial displacement u [mm] 8 6

Qmax: 250 N 200 N

8 6

400 N

190 N

330 N

4 2

4
320N

175 N

2 0

0
(a)

0.02 Time t [s]

0.04

0.06
(b)

0.02 Time t [s]

0.04

0.06

10 Axial displacement u [mm] 8 6 4 2


660N 600N

10 Axial displacement u [mm]


Qmax: 700N

8 6 4 2

Qmax: 1200 N

680N

1100 N

1075 N

1050 N

0
(c)

0.02 Time t [s]

0.04

0.06
(d)

0.02 Time t [s]

0.04

0.06

10 Axial displacement u [mm] 8 6 4 2


1700 N Qmax: 2000 N 1800 N

1750 N

0
(e)

0.02 Time t [s]

0.04

0.06

Fig. 2. Axial displacement of column. Axial load P 275 N and transverse-impact position xQ 160 mm. (a) Transverse-impact duration T 1.0 ms, (b) T 0.5 ms, (c) T 0.2 ms, (d) T 0.1 ms, and (e) T 0.05 ms.

ARTICLE IN PRESS
A. Sastranegara et al. / Thin-Walled Structures 44 (2006) 701707 703

We previously studied the effect of transverse-impact on a statically compressed column in an experiment [4]. In this paper, we discuss the conditions for transverse-impact to generate static-column buckling and the effect of the transverse-impact on post-buckling behavior with niteelement analysis. 2. Analysis The model used for numerical analysis was a column clamped at the bottom and simply supported at the top end, as outlined in Fig. 1. The column was statically compressed by axial load P. Under axial compression, transverse impact was applied to the column. The transverse load-history Q(t) was approximated as a halfsine shape as described by Eq. (1), based on the experimental results.   Qmax 2pt Qt 1 cos 0otoT ; 0T ot, T 2 (1)

where Pcr, I, and L are Eulers buckling load, the second moment of the area, and the length of the column, respectively. Therefore, this imperfection was sufciently small and did not affect the buckling load. The loading process in this analysis was done as follows. First, axial load was quasi-statically applied at the simply supported top end. The load was kept constant after reaching the setting value. Transverse-impact was then applied to the column with the parameters listed in Table 1.

10

8 Axial displacement u [mm] Qmax: 525 N 6 500 N

where Qmax and T are the maximum and the duration of the transverse-impact load. The material for the column had perfectly elasto-plastic properties. The elastic modulus E, yield stress sy , and density r of the column were 56 GPa, 60 MPa, and 2700 kg/m3, respectively. The column was 400-mm long, 20-mm wide, and 3-mm thick. The histories of impact load Q(t) and impact position xQ were varied under different static axial loads. The parameters used in the numerical analysis are listed in Table 1. Finite-element analysis was done with an FEM package, MARC, version K7. The column was discretized by 400 Timoshenkos beam elements with three nodal points to obtain sufcient accuracy even for post-buckling behavior. Two displacements and one rotation were admissible at each node. Three-point Gaussian integration for mass and two-point integration for stiffness were used. These are the most commonly used elements in dynamic problems, because of the importance of shear deformation and rotational inertia on column response at high frequency. In the analysis, both large displacement with small strain and geometrical non-linearity based on total Lagrangian formulation were taken into consideration. The Newmarkb time integration technique was used to solve the dynamic response of the column. An initial imperfection is generally applied to a structure model in buckling analysis. The imperfection should be sufciently small to have no effect on buckling load and buckling behavior. In our analysis, the initial imperfection had the same shape as the rst mode for bending vibration with amplitude of 0.1% of the column thickness. The buckling load computed for the rst mode without transverse load was 317 N. This was very close to Eulers buckling load, 318 N, given by Eq. (2) [14] Pcr 20:19 EI , l2 (2)

475 N

450 N 0 0.02 Time t [s]


Fig. 3. Axial displacement of column. Axial load P 300 N, transverseimpact duration T 0.2 ms, and transverse-impact position xQ 160 mm.

0.04

0.06

10

8 Axial displacement u [mm] Qmax: 300 N 6 220 N 4 260 N

180 N

0.02 Time t [s]

0.04

0.06

Fig. 4. Axial displacement of column. Axial load P 315 N, transverseimpact duration T 0.2 ms, and transverse-impact position xQ 160 mm.

ARTICLE IN PRESS
704 A. Sastranegara et al. / Thin-Walled Structures 44 (2006) 701707

3. Results Fig. 2 plots typical histories of displacements at the top end of the column, u under axial load P 275 N, for different Qmax and impact durations T. The solid and the broken lines denote buckling and unbuckling cases, respectively. In Fig. 2(a), the displacement when Qmax 175 N increased rapidly just after transverseimpact, reached a maximum, and gradually decreased. The nal displacement was due to plastic bending deformation in the column. However, the displacements at Qmax above 190 N continually increased due to the generating of buckling. As transverse-impact duration T became shorter, a larger Qmax was required to collapse the column. The buckling in Fig. 2(a) occurred at approximately 0.01 s when near critical Qmax (190 N) was applied to generate buckling, while it occurred in Fig. 2(e) at approximately 0.03 s when near critical Qmax (1750 N) was applied. It was interesting

to note from Fig. 2 that the shorter transverse-impact duration T took the column a longer time to buckle at the critical Qmax. Axial displacement before buckling was larger if the duration of transverse-impact was longer. This meant that large bending deformation in the column occurred before buckling. Figs. 3 and 4 show the displacement histories for axial loads P 300 and 315 N at transverse-impact duration T 0.2 ms. It was obvious from these gures that for the higher axial load, the critical Qmax decreased. The results in Figs. 24 indicate that the generation of buckling is dependent on both Qmax and T. Fig. 5 shows the deformation histories of the column for different positions of transverse-impact, xQ, until buckling occurs. Axial static load P is 275 N and impact duration T is 0.2 ms. Although a different high mode for vibrations for each impact position appeared just after impact, every column nally deformed with the rst mode for static buckling.

Fig. 5. Deformation histories of column. Transverse-impact duration T 0.2 ms, and axial load P 275 N. (a) Transverse-impact position xQ 80 mm and maximum transverse-impact load Qmax 735 N, (b) xQ 160 mm and Qmax 680 N, (c) xQ 240 mm and Qmax 660 N, (d) xQ 320 mm and Qmax 660 N.

ARTICLE IN PRESS
A. Sastranegara et al. / Thin-Walled Structures 44 (2006) 701707 705

4. Discussion Fig. 6 denotes the relation between transverse-impact duration T and maximum transverse-impact load Qmax for axial load P 275 N and transverse-impact position xQ 160 mm. The critical maximum for transverse-impact load to buckle of the column was larger for the shorter impact duration. Fig. 7 shows the effect of transverse-impact duration on critical transverse impulse M in generating buckling. Transverse impulse M was dened as Z M
0 T

120

Critical transverse impulse M [Ns]

90

60

Qtdt.

(3)

30

From this gure, we can see that the critical impulses were dependent on the durations of transverse-impact. Consequently, the transverse impulse was not a parameter governing the generation of buckling. Fig. 8 shows the critical transverse-impact energy to buckle the column with transverse-impact duration for axial load P 275 N and transverse-impact position xQ 160 mm. Transverse-impact energy Etr was dened as work by the transverse impact: Z w E tr Qdw, (4)
0

0.25 0.50 0.75 Transverse impact duration T [s]

1.00

Fig. 7. Relation between critical transverse-impulse M and transverseimpact duration T. Axial load P 275 N and transverse-impact position xQ 160 mm.

0.5

Transverse impact energy Etr [J]

where Q and w are transverse-impact load and deection of the column at impact point xQ, respectively. In this gure, the critical energy of transverse-impact to buckle was approximately constant regardless of impact duration T or maximum transverse-impact load Qmax. Then, the major parameter that induced buckling in an axially compressed column was claried to be transverse-impact energy.
2000 Maximum transverse impact load Qmax [N] Buckling Unbuckling 1500

0.4

0.3

0.2

Critical energy of transverse impact 0.1 Buckling Unbuckling

0 0.01

0.05 0.10 0.50 Transverse impact duration T [ms]

1.00

1000 Critical line


Fig. 8. Critical energy of transverse-impact for different transverse-impact durations T. Axial load P 275 N and transverse-impact position xQ 160 mm.

500

0.25 0.50 0.75 Transverse impact duration T [ms]

1.00

Fig. 6. Critical maximum of transverse-impact load for different transverse-impact durations T. Axial load P 275 N and transverseimpact position xQ 160 mm.

The relation between axial load and critical transverseimpact energy for transverse-impact position xQ 160 mm is plotted in Fig. 9. The critical energy linearly decreased with axial load. Even though transverse-impact duration was different, the critical energy of transverse-impact was approximately the same. This again proves that the governing parameter was transverse-impact energy. These results agree with the experimental results [4].

ARTICLE IN PRESS
706 A. Sastranegara et al. / Thin-Walled Structures 44 (2006) 701707

0.4

5
A

Critical transverse impact energy Etr [J]

0.3

Axial velocity du/dt [m/s]

T = 1.0 ms T = 0.2 ms T = 0.1 ms

A: P = 275 N, Qmax = 700 N B: P = 300 N, Qmax = 525 N C: P = 315 N, Qmax = 300 N D: P = 300 N, Qmax = 500 N E: P = 315 N, Qmax = 260 N F: P = 275 N, Qmax = 680 N

D B C E F

0.2 Static buckling load 0.1

0
0 270 280 290 300 Axial load P [N] 310 320

0.01

0.02 Time t [s]

0.03

0.04

Fig. 9. Relation between critical transverse-impact energy Etr and axial load P. Transverse-impact position xQ 160 mm.

Fig. 11. Histories of velocity at the top of column under various conditions. Transverse-impact duration T 0.2 ms and transverse-impact position xQ 160 mm.

0.8 T = 1.0 ms T = 0.2 ms T = 0.1 ms

Critical transverse impact energy Etr [J]

0.6

column can buckle more rapidly. After buckling, the acceleration at the top end of the column converges to the same value, which is indicated by the same slopes of these curves. 5. Conclusion

0.4

0.2

100 200 300 Transverse impact position xQ [mm]

400

Fig. 10. Relation between critical transverse-impact energy Etr and transverse-impact position xQ.

Fig. 10 denotes the critical impact energy for different positions of transverse-impact. When a column was transversely impacted in its middle area (xQ 160 to 320 mm), the energy required for buckling was relatively constant regardless of the duration of impact. Fig. 11 shows the histories of axial velocities du/dt at the top of the column for different axial loads P and maximum transverse-impact loads Qmax. Buckling occurred in all cases of this gure. If maximum transverse-impact load Qmax becomes higher with the same axial load P, the

The effect of transverse-impact on the buckling behavior of a column under axial compressive load was numerically investigated with nite-element analysis. An axially compressed column was subjected to transverse-impact that instantaneously reduced its structural stiffness. The numerical results revealed that transverse-impact had a signicant effect on static buckling. A column buckled under statically axial load lower than Eulers axial buckling load due to transverse-impact. The generation of buckling was governed by transverse-impact energy. The duration of transverse-impact duration had a signicant inuence on its critical maximum load of transverseimpact. The shorter the duration of impact, the higher the critical transverse-impact load that was required to generate buckling. The velocity of column deformation was found to increase by increasing maximum transverse-impact load. Therefore, deformation during the post-buckling process may be able to be controlled by adjusting transverseimpact. An axially compressed column subjected to transverseimpact buckles in low mode. However, shortly after transverse-impact, the column vibrated at higher modes. These results agreed qualitatively with the results of the experimental investigation [4].

ARTICLE IN PRESS
A. Sastranegara et al. / Thin-Walled Structures 44 (2006) 701707 707

References
[1] Ezra AA, Fay RJ. An assessment of energy absorbing devices for prospective use in aircraft impact. In: Herrmann G, Perrone N, editors. Dynamic response of structures. New York: Pergamon; 1972. p. 22546. [2] Macaulay M. Introduction to impact engineering. London: Chapman & Hall; 1987. p. 14072. [3] Jones N. Structural impact. Cambridge: Cambridge University; 1989. p. 40815. [4] Adachi T, Tanaka T, Sastranegara A, Yamaji A, Kim SK, Yang IY. Effect of transverse impact on buckling behavior of a column under static axial compressive force. Int J Impact Eng 2004;30:46575. [5] Sastranegara A, Adachi T, Yamaji A. Improvement of energy absorption of impacted column due to transverse impact. Int J Impact Eng 2005;31:48396. [6] Sastranegara A, Adachi T, Yamaji A. Improving energy absorption of impacted column due to transverse impact: nite element analysis. Int J Impact Eng 2005;32:44460.

[7] Libove C. Elastic stability. In: Flu gge W, editor. Handbook of engineering mechanics. New York: McGraw-Hill; 1962. p. 44-142. [8] Johnson W. Impact strength of material. New York: Edward; 1972. p. 2718. [9] Jones N. Structural impact. Cambridge: Cambridge University; 1989. p. 909. [10] Cost TL, Jones HW. Dynamic response of blast loaded pre-stressed at plates. J Sound Vib 1979;62:11120. [11] Zeinoddini M, Harding JE, Parke GAR, Harding JE. Dynamic behavior of axially preloaded tubular steel members of offshore structures subjected to impact damage. Ocean Eng 1999;26:96378. [12] Chen FL, Yu TX. Inuence of axial pre-load on plastic failure of beams subjected to transverse dynamic load. Key Eng Mater 2000;177180:25560. [13] Zeinoddini M, Parke GAR, Harding JE. Axially pre-loaded steel tubes subjected to lateral impacts: an experimental study. Int J Impact Eng 2002;27:66990. [14] Timoshenko S, Gere JM. Theory of elastic stability. 2nd ed. New York: McGraw-Hill; 1961. p. 137.

Vous aimerez peut-être aussi