Proceedings of the 9th WSEAS International Conference on Applied Mathematics, Istanbul, Turkey, May 27-29, 2006 (pp128-133)

Rollover Analysis of a Bus Using Beam and Nonlinear Spring Elements

SU-JIN PARK, WAN-SUK YOO, YUEN-JU KWON
School of Mechanical Engineering
Pusan National University
30 Jangjeon-dong Geumjeong-gu Busan, 609-735
SOUTH KOREA
http://cae.me.pusan.ac.kr

Abstract: - In case of bus rollover, the body structure of a bus should be designed to ensure the survival space for
passengers. So, this study focused on evaluating the rollover strength through a computer simulation using the
commercial code, LS-DYNA3D at the initial stage of vehicle development. For this purpose, section structure of
the bus frame was first modeled using simple beam elements, and the impact boundary conditions required by ECE
regulation No. 66 were applied. In order to validate the beam element model, the simulation results were
compared to those of shell element model. Since the analysis error with beam element model was mainly caused
by the difference in strain energy at the T-shaped connecting joint, the joint connection was modified with
nonlinear springs to account for the local buckling. With the improved beam model, the results were in a good
agreement to shell element model, but the simulation times were much reduced.

Key-Words: - Rollover strength, Survival space, Beam element, Nonlinear spring, Collapse theory

1 Introduction developed by Richardson [5] and Larson [7], and

Many types of buses have been used for Moffatt [8] used the controlled rollover impact
commercial traffic transportation, and life span of a system for rollover impact investigation. In this
bus is longer than that of a passenger car. Also, a study, the commercial code LS-DYNA3D [11] was
bus have been operated with much more passengers used for the rollover analysis of bus structure, and
than a small passenger car, however, the research the beam analysis model was modified with
and interest related on its safety seemed relatively nonlinear spring characteristics for the advanced
low. Since there are many seriously wounded exactness of its solution.
passengers in traffic accidents of buses, more
careful analysis and simulation for impact accidents
should be investigated. Since the understanding and 2 Rollover Test
knowledge of automobile consumers for their safety The rollover test of real section structure was
had been growing up, the related official rules have executed by a bus manufacturer [15] according to
come into effect. In many bus manufacturers and the conditions which were defined by ECE. The
research institutes, new designs and various body section was located on a rotational platform,
researches are in progress to match new regulations. and the platform was lifted up by a crane as shown
In Australia, ADR 59 (Australian Design Rule in Fig. 1. The body section structure has the weight
59) had specified the deformation of bus body in of 830 kgf, and the free dropping begins after the
rollover accidents [1], and ECE 66 (Economic center of gravity moves out of the hinge point.
Commission for Europe 66) controls the same As the results of rollover test, a deformed body
substance presently [2]. Bus structure designs that section is shown in Fig. 2. Plastic deformation
ensure the survival space of passengers are occurred near the welding point of side body beams,
indispensable, and for these purposes, the rollover and also the local break-down appeared on the open
analysis and design by using explicit finite element section of the beams. The bending collapse took
method are useful [3,4]. Takagi [6] has numerically place at the joining area of lower part of window
simulated rollover occurrence and occupant frame and side beams as shown in Fig. 3. The
behavior according to the rollover initiation types strength of this area dominates deformation of the
with commercial computer program. The physical whole body, and this connection is the most
test method for rollover accident has recently
Proceedings of the 9th WSEAS International Conference on Applied Mathematics, Istanbul, Turkey, May 27-29, 2006 (pp128-133)

important to ensure the survival space of passengers Fig. 4 Rollover analysis of shell element model
in rollover accidents.
3 Numerical Rollover Simulation

3.1 Analysis with Shell Element Model

The body section structure is a skeleton set which
is composed of frame-type pillars. This structure is
a framework of the whole body which absorbs
60~70 % of the impact energy when impact takes
place [9,10]. The exterior panels and leads were
neglected for their small capacity of impact energy
absorption. Shell elements were used to establish
Fig. 1 Rollover test apparatus the body section structure as shown in Fig. 4, and
concentrated mass elements were arranged to adjust
the weight of the structure. The material properties
of elasto-plastic steel were employed, and the
nonlinear region of material property curve was
piece-wise linearly approximated.

3.2 Analysis with Beam Element Model

In the design stage of a bus, the numerical
evaluation of rollover features with a simple beam
model has the advantages of its low cost and quick
Fig. 2 Rollover test of the body section structure
response [12]. Moreover, the reduction of effort to
generate the analysis model is also possible. The
Belytschko-Schwer beam elements [11] with the
material properties of resultant plasticity, which was
adequate for box-tube type structures, were selected
as shown in Fig. 5. The cross-sectional area,
moment of inertia, torsional modulus, and the
properties of shear area were considered without
discrimination with the previous shell model [11].
Also, the concentrated mass elements and the same
boundary conditions were applied.

Fig. 3 Side pillar frame under bending collapse

Fig. 5 Rollover analysis of beam element model

3.2 Comparison of the Two Results

Proceedings of the 9th WSEAS International Conference on Applied Mathematics, Istanbul, Turkey, May 27-29, 2006 (pp128-133)

It was cleared that the survival space for

passengers was secured via computer simulations. Fig. 6 Comparison of absorbed energy
Deformation of window pillars were similar to the
results of body section test, and the survival spaces 4 Simulation with Modified Models
were obtained during rollover procedures as shown
in Fig. 4 and Fig. 5. Nonlinear spring elements were supplemented to
The displacement of window pillars measured improve the disadvantages of beam element bus
from the rain rails and deformation angles from the model. Nonlinear spring elements were added at the
waist rails are summarized in Table 1. The joining area of beam elements to consider the local
displacement and deformed angle in beam element buckling and the adequate structural stiffness.
model were smaller than those of shell element
model, and the differences were 9 % and 12 % of the 4.1 Collapse Theories
results in shell element model, respectively. The scheme for nonlinear spring input is
The strain energy absorbed by the body section in described in Fig. 7. The characteristics for axial
deformation processes is shown in Fig. 6. The crush, bending collapse, and torsional collapse are
absorbed energy of beam element model was higher required to import nonlinear spring elements, and
than that of shell element model. These differences the curves of force-displacement,
come from the over-estimated stiffness of beam moment-rotational angle, and torsion-twisting angle
element model. In the beam element model, no local should be computed in advance to the simulations,
buckling at the area of plastic hinge was considered, respectively. The following equations were adopted
and as the result, plastic deformation was not to represent the axial crush and the average crush
sufficiently estimated. Therefore the beam element force [13].
modeling with the consideration of the stiffness of
pillar joint area is needed to obtain more exact 1
a + b 3
simulation results. Pasm = 16 m p , (1)
2t
Table 1 Comparison of displacements and y t2
deformed angles measured at window pillar mp = , (2)
Test 392 mm 4
Shell element Padm = f d Pasm , (3)
397 mm
Displacement model f d = 1 + 0.03 v0
0.256
. (4)
Simulation
Beam element
363 mm
model
Test 24.9 The maximum bending moment and
Shell element post-buckling behavior are important factors to
Deformed 25.6 describe the bending collapse of structural frames.
model
Angle Simulation
Beam element The relations on the bending collapse defined by
22.5
model Kecman and Wierzbicki were used as followings
[14,17].

M 0b rb
M ( ) =
H b
3500 beam model 8 I1 + 2I 2 b + 2
shell model sin t rb Hb
3000

2 b Hb
2
2500
1 +
Absorbed energy (J)

+
2000 2 Hb
b / 2

( )
1500
b H
+ 4 sin + 2 2 b , (5)
t (b )
1000
b
500 2 1

0
H b = 1.276 b 3 t 3 , (6)
1 2
-500
0.0 0.5 1.0 1.5 2.0 2.5 3.0
rb = 0.795 b t , 3 3
(7)
Time (sec)
Proceedings of the 9th WSEAS International Conference on Applied Mathematics, Istanbul, Turkey, May 27-29, 2006 (pp128-133)

1
I1 = cos 0
1 + cos 2
d , (8)

I 2 = cos 1 + sin 2 , (9) Fig. 8 Torsion-twisting angle curve

tan = tan ( 2 . ) (10)

In the joint area, the eccentric loading is applied,

and torsional deformation occurs. The behavior of
thin-walled beams under the torsional loading is
effectively treated using initial twisting moment Ti ,
twisting moment in steady-state Ts , critical twisting
angle c , and the final twisting angle f as shown
in Fig. 8, and the following equations were used to
calculate the torsional energy EA [16]. Fig. 9 Modified model for T-shaped joint

EA Ts f
c = , (11) 6
Ti Ts
Ti = 3.72 y t 1.25 b1.75 ,
5
(12)
Ts = 5.58 0 t 1.4 b 2.2 l 0.6 , (13) 4 beam model
shell model
Force (kN)

3 beam-nonlinear spring model

where 0 denotes the stress value that corresponds to

2
the average strain.
1

0
0 2 4 6 8 10 12

Time (msec)

Fig. 10 Reaction force curves against bending

3000
Torsional moment (Nm)

2500
beam model
shell model
2000
beam-nonlinear spring
model
1500

Fig. 7 Scheme for nonlinear spring input 1000

500

0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8

Rotetional angle (radian)

Fig. 11 Reaction torsion curves against twisting

4.2 Verification with Simple Joint Model

Proceedings of the 9th WSEAS International Conference on Applied Mathematics, Istanbul, Turkey, May 27-29, 2006 (pp128-133)

To verify the effect of nonlinear spring be reasonable in the previous section. The joining
attachment, T-shaped simple joint model was areas between window pillars and side beams, at
investigated as shown in Fig. 9. Nonlinear springs which the local buckling and plastic hinge had
were located on the joining point and its both sides. originated in the previous test and simulations, were
As previously mentioned, the collapse conditions by remodeled as shown in Fig. 14.
Wierzbicki and Santosa were applied to activate the The resultant deformed shape from the rollover
added spring elements [13,16]. simulation with the modified bus model is in Fig. 15.
Reaction properties against the bending and Bending at the modified joint area was increased,
torsional load are plotted in Fig. 10 and Fig. 11, and and the resultant difference was reduced to 3 % of
the absorbed energy curves are in Fig. 12 and Fig. 13, shell element bus model. The absorbed energies are
respectively. It was verified that the modified model plotted in Fig. 16, and this comparison shows the
represented well the nonlinear characteristics of improved result which contains the difference of
joint area. 1.2 % in their energy values.

1000
beam model
shell model
800
beam-nonlinear spring
Internal energy (kJ)

model

600

400

200

0
0 2 4 6 8 10 12

Time (msec)
Fig. 14 The locations of modified joints

Fig. 12 Absorbed bending energy curves

beam model
1000 shell model
beam-nonlinear spring
model
800
Internal energy (kJ)

600

400

Fig. 15 Deformed shape of modified bus model

200

beam model
3500 shell model
0
0 2 4 6 8 10 12 14 16 18 beam-nonlinear
3000 sproing model
Time (msec)
2500
Absorbed energy (J)

Fig. 13 Absorbed twisting energy curves 2000

1500

1000
4.3 Modified Bus Model and Improved
500
Results
The beam element bus model was modified with 0

the nonlinear spring elements which were verified to -500

0.0 0.5 1.0 1.5 2.0 2.5 3.0
Time (sec)
Proceedings of the 9th WSEAS International Conference on Applied Mathematics, Istanbul, Turkey, May 27-29, 2006 (pp128-133)

and Occupant Behavior in Each Rollover Initiation

Fig. 16 The comparison of absorbed energy Type, SAE 2003-06-0071, Nagoya, Japan, 2003.
[7] R. E. Larson, J. W. Smith, S. M. Werner, G. F.
5 Conclusion Fowler, Vehicle Rollover Testing, Methodologies
In the rollover simulation of a finite element bus in Recreating Rollover Collisions, SAE
model, a beam element model was successfully 2000-01-1641, Troy, Michigan, 2000.
modified to save the analysis cost with little loss of [8] E. A. Moffatt, E. R. Cooper, J. J. Crotean, K. F.
its accuracy in comparison with a shell element Orlowski, D. R. Marth, J. W. Carter, Matched-Pair
model. The stiffness of simple beam elements was Rollover Impacts of Rollcaged and Production
overestimated, and the local plastic deformation was Roof Cars Using the Controlled Rollover Impact
partially out of consideration. However, the System, SAE 2003-01-0172, SAE World Congress,
modified model, to which spring elements and its Detroit, Michigan, 2003.
nonlinear characteristics were applied, estimated the [9] H. W. Lee, H. W. Shin, A Study on the
structural stiffness properly, and showed the Characteristics of Frontal and Oblique Impact of a
improved results of reaction forces and internal 4WD Vehicle, Autumn conference proceeding of
energy. The stability and effectiveness of joint KSAE, 1997, pp. 441-446.
modification technique was confirmed, and the [10] C. S. Hong, C. Y. Oh, D. C. Lee, Development of
approach was proved to be useful and advantageous a Finite Element Model for Frontal Crash Analysis
in the vehicle rollover investigation and its of a Mid-Size Truck, Journal of KSPE, Vol. 17,
development. No. 4, 2000, pp. 226-232.
[11] Livemore Software, LS-DYNA THEORETICAL
MANUAL, 1998.
Acknowledgements: [12] H. K. Min, T. Kim, An Analysis for Rollover
This study was financially supported by Pusan Strength of a Medium Bus, SAE NO.99370235,
National University in the program, Post-Doc. 2006. 1999, pp. 195-201.
The authors thank the Ministry of Science and [13] T. Wierzbicki and W. Abramowicz, On the
Technology of Korea for its sincere support under Crushing Mechanics of Thin-Walled Structures,
the NRL project (M10400000274-05J000027410). ASME Journal of Applied Mechanics, Vol. 50,
1983, pp. 720-739.
[14] D. Kecman, Bending Collapse of Rectangular
References: and Square Section Tubes, Int. J. of Mech., Vol.
[1] ADR 59, Australian Design Rule 59/00, Omnibus 25, No. 9-10, 1983, pp. 623-636.
Rollover Strength. [15] Technical Report, E 130, The Analysis and
[2] United Nation, Addendum 65: Regulation No.66, Experiment on the Rollover Accident of a Bus,
Uniform Provision Concerning the Approval of Daewoo Bus Corporation.
Large Passenger Vehicles with regard to the [16] S. Santosa and T. Wierzbicki, Effect of an
Strength of their Superstructure, 1996. Ultralight Metal Fillar on the Torsional Crushing
[3] T. E. Chung, Rollover Analysis and Measurement Behavior of Thin-Walled Prismatic column, Int. J.
of a Large-sized Bus, Transaction of KSAE, Vol.5, Crashworthiness, Vol. 2, No. 4, 1997, pp.
No.6, 1997, pp. 148-154. 305-331.
[4] H. D. Kim, J. H. Song, C. Y. Oh, Development of [17] S. H. Park, S. Y. Kang, H. Y. Kim, D. K. Min, D.
a Finite Element Model for Crashworthiness C. Han, Development of Nonlinear Spring
Analysis of a Small-Sized Bus, Transaction of Element and Vehicle Crashworthiness Analysis,
KSAE, Vol.10, No.1, 2002, pp. 153-161. Proceedings of KSME Conference, 97380165,
[5] S. Richardson, R. H. Grzebieta, G. Rechnitzer, 1997, pp. 407-412.
Proposal for a Dynamic Rollover Protective
System Test, International Journal of
Crashworthiness, Vol. 8, No. 2, 2003, pp.
133-142.
[6] H. Takagi, A. Maruyama, J. Dix, K. Kawaguchi,
MADYMO Modeling Method of Rollover Event