See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/237762573

MOTORCYCLE-CAR SIDE IMPACT SIMULATION

Article

CITATIONS READS

20 664

4 authors, including:

Sudipto Mukherjee A. Chawla

Indian Institute of Technology Delhi Indian Institute of Technology Delhi
178 PUBLICATIONS   781 CITATIONS    124 PUBLICATIONS   1,032 CITATIONS   

SEE PROFILE SEE PROFILE

Dinesh Mohan
Indian Institute of Technology Delhi
187 PUBLICATIONS   5,657 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Road Traffic accident data collection and crash analysis View project

impact biomechanics View project

All content following this page was uploaded by Sudipto Mukherjee on 25 August 2014.

The user has requested enhancement of the downloaded file.

Back

Back
MOTORCYCLE-CAR SIDE IMPACT SIMULATION

S. Mukherjee1, A. Chawla1, D. Mohan1, M. Singh1

M. Sakurai2, Y. Tamura2

1 Indian Institute of Technology Delhi

Transportation Research & Injury Prevention Programme
Hauz Khas, New Delhi 110016, India

2 Japan Automobile Research Institute, Japan

ABSTRACT

[ISO13232] specifies four configurations of motorcycle-car side impact tests. In this study a
Kawasaki GPZ motorcycle and Toyota Corolla were used to conduct crash tests according to ISO
specifications at the Japan Automobile Research Institute (JARI). Finite element models of the
motorcycle and car were developed in PAM/CRASH™ based on CMM measurement and component
test data. In this paper we present the results of the simulation of four ISO standard crash
configurations. The kinematics in these simulations is compared with those in the experiments.

KEY WORDS
COMPONENT TESTS, HYBRID III, DUMMIES, SIDE IMPACT, ISO, KINEMATICS,
PAMCRASHTM

IN MANY ASIAN COUNTRIES motorcycle (M/C) crashes constitute a significant proportion of all
road crash fatalities and injuries and safety measures can be successful only if much more attention is
devoted to this issue. In this paper we investigate car-motorcycle side impact simulations which have
been compared and validated using crash tests conducted earlier.

Earlier experimental as well as simulation based studies to evaluate the effectiveness of Leg
Protectors (LPs) for motorcyclists have been carried out by [Chinn,1986]. Studies have also been
presented on the modeling of airbags in motorcycles [Nieboer,1991]. Subsequently, [Nieboer,1993]
have reported development of rigid body models of the motor cycle. We concur with their observation
that , MC simulations turn out to be far more difficult compared to the simulation of car occupants.
This is due to the multiplicity and complexity of the interactions involved. Use of rigid body models
in these simulations have the drawback that the energy absorption is underestimated specially in cases
of large deformations [Nieboer,1993]. As a result we have initiated research using FE based tools to
understand the important issues in the crash behaviour of MCs. This paper describes initial work in
that direction.

The ISO 13232 specifies seven crash configurations, four of which are side impacts. In this
paper we present the results of crash tests simulations involving a Kawasaki GPZ motorcycle (with a
dummy rider) and a Toyota Corolla. The crash tests were conducted at the Japanese Automobile
Research Institute (JARI) and the computer simulations at TRIPP, Indian Institute of Technology,
Delhi. Simulations of four side impact situations have been developed. The study has been sponsored
by the Japan Automobile Manufacturer’s Association (JAMA).

The Kawasaki GPZ, 1990, 400 cc model, the Toyota Corrola, 1989 model and the MATD
dummy were used in the experimental tests. Simulation models for the vehicles were developed using
geometric data obtained from a Coordinate Measuring Machine (CMM). This includes surfaces like
M/C headlight, seat, tank, engine, side panels and leg protectors, car bonnet, bumper, side panel. This
data was used to develop the geometric model of the motorcycle in IDEAS. The modelled
components were further verified against ISO standard static tests in IDEAS and impact tests by
porting to PAM/CRASH. Simulation of the full motorcycle assembly using PAM/CRASH was
verified against an experimental wall crash. Sufficient confidence has been established for the
behaviour of the motorcycle model in impact conditions. The H-III dummy model available in
PAM/CRASHTM was used. The current work concentrates on the kinematics and the dynamics of the
M/C rider. The details of the methodology followed are described in [Chawla, 2001]. The detailed
finite element modelling of the M/C has provided a relatively reliable tool for assessing motorcycle-
car crash details under differing crash configurations.

METHODOLOGY

The exact geometric data in the form of CAD model or drawings and constitutive properties
of the material used in car and motorcycle components were not available. A coordinate measuring
machine (CMM) was used to collect geometric data of components. Static and dynamic testing of the
components were conducted at JARI to measure the material properties of components like bumper,
bonnet, Motor Cycle (M/C) seat foam, front suspension and tire, to name a few. In dynamic tests,
components were impacted at speeds set in the ISO standards. Further, to get the static material
properties of some materials, load deflection characteristics were recorded using an Ultimate Tensile
Machine (UTM).

Finite element models were developed for components of the car and motorcycle. For
geometric definition and mesh generation, IDEASTM was used. Static validation of component
models against experimental data was also carried out in IDEASTM. The mesh was then exported and
assembled in PAM/CRASHTM in order to validate and build the complete model of the car and the
motorcycle under impact conditions. Material properties and the boundary conditions were defined in
PAM/CRASHTM. Experimentally recorded curves of displacement, velocity and acceleration were
compared against the values obtained by simulation. This helped in selecting the material properties
of components where properties were not known otherwise and gave confidence in their fidelity
[Chawla, 2001].

While we used FE models of the car and the MC, we have used a rigid body model for the
dummy. The dummy model used is the H-III dummy model available with PAM/CRASHTM. The
dummy used in the car – motorcycle tests is an MATD dummy. This change does affect the
simulations. However, in this paper we are looking primarily at the overall kinematics of the car – MC
impact and the overall MC behaviour would not change much. A more appropriate dummy model will
however have to be used for looking at the finer aspects of the simulations.

FINITE ELEMENT MODEL OF THE CAR

As mentioned above, the initial modeling of the car components was done in IDEASTM to
capture the geometry. Component models were then transported to PaM/Crash and the complete car
model was assembled. For meshing, four node shell elements and beam elements were used. The
elements on the impacting side of the car were more refined (smaller) compared to the rest of the car.
This was to done to minimize the simulation time. The complete car model consists of 14828 Shell
elements, 432 brick elements, 465 beam elements, 4 kinematic joints and 4 spherical Joints.

For the side impact of the car with motorcycle, the components of the impacting side of the
car, the front and rear door, side panel and B-pillar were modeled with dense mesh. The non-
impacting side of the car, far away from the impact zone, the bonnet, bumper, engine, radiator,
windscreen, tires of the car were initially defined as rigid null material so that the internal forces and
deformation are not computed during the simulation to save on simulation time. Subsequently,
appropriate properties are given to these parts as well.

The front door of the car was modeled using the 4730 four node thin shell elements and 148
beam elements. Similarly the rear door of the car has been modeled using 4421 shell elements and
134 beam elements. The side panel of the car consisting of A-pillar, B-pillar, the roof rail and the
bottom rail has 1955 shell elements.

The chassis of the car has initially been modelled using beam elements. During the impact
with the motorcycle the chassis of the car has little or no permanent deformation. Consequently
equivalent beam elements were used to speed up the computation.

The suspension has been modelled as translating kinematic joints, whose stiffness and
damping characteristics were determined. The spin of the wheel has been modelled as a spherical
joint element constrained to have free rotation only about the wheel axis. An initial linear velocity is
assigned to the entire car when needed. The corresponding angular velocity is defined for the car
tires. Contact interactions and friction are defined between contacting surfaces like tires and road,
doors and frames, inside and outside surfaces of door, to name a few. Mass and moment of inertia of
the car was made equivalent to that of the car by introducing point masses at appropriate locations.
Exact geometric models were not built for the engine, the transmissions and seats as these do not have
a significant contribution to the car-M/C impact.

FINITE ELEMENT MODEL OF THE MOTORCYCLE

The geometry of the motorcycle was also reproduced in IDEASTM software a component at a
time. Components were validated and assembled in PAM/CRASHTM. Particular attention was given to
reproduce the front suspension characteristics and the rear suspension mechanism as they play a
significant role in determining the trajectory of the dummy in impact situations. The motorcycle
model currently consists of 96 solid elements, 850 shell elements, 102 beam elements, 1 bar element,
2 kinematic joints and 9 spherical joints.

For the car tire, the geometry has been preserved. The load deformation characteristics of the
tire have been clubbed together with the suspension. However for the motorcycle, fidelity of the tire
is important, as it is the initial impact body. The motorcycle tire is a shell mesh with material
characteristics of rubber. An internal tire pressure of 200 KPa was defined for the front and rear tire
of the motorcycle. The model has been built in accordance with the methodology validated by
[Takatori, 1999].

The finite element model of the motorcycle was validated using the experimental data of a
motorcycle-wall crash. For the validation, the wall forces and the acceleration of the CG point and
point under the seat were compared.

The motorcycle dummy used was the 50th % Hybrid III rigid body model available with
PAM/CRASHTM. The dummy segments were reoriented for it to sit on the motorcycle. The MATD
dummy used for the experimental tests has joint stiffness that differ from that of the Hybrid III. The
affect of some of these differences is being discussed in subsequent sections.

SIMULATION ENVIRONMENT OF CAR MOTORCYCLE SIDE IMPACT

In the ISO 13232 the four different configurations of the side impact of car motorcycle impact
are:

Case I: Car stationary, motorcycle impacting at velocity of 13.4m/s at right angle

Case II: Car moving at 6.7m/s, motorcycle moving at 13.4m/s at angle from of 450 from front.
Case III: Car moving at 6.7m/s, motorcycle impacting at a velocity of 13.4m/s at right angle
Case IV: Car moving at 6.7m/s, motorcycle moving at 13.4m/s at angle from 450 from rear.
Fig.1 : Top view of M/c & Car impact Fig. 2 : Top view of M/c & Car impact
configuration for – Case I configuration for – Case II

Fig. 4 : Top view of M/c & Car impact

Fig. 3 : Top view of M/c & Car impact configuration for – Case IV
configuration for – Case III

In all the configurations the motorcycle is to impact at the center of the car i.e. at L/2 of the
car where L is the length of the car. The four different configurations are shown in the figures [1-4]

Contact interactions were defined between the dummy hip, thighs, legs and foot and the
complete motorcycle. Contact interactions were also defined between the forward segment of the
motorcycle and the side of the car. The reader may recollect that the side of the car near the impact
area had refined mesh geometry for good fidelity. A helmet model, verified through drop test was
assembled on to the head of the dummy. Contact interactions were defined between the head and the
helmet and between the helmet and the car. Figure 5 shows the complete model of the car motorcycle
side impact. IN the next section we describe the kinematics in these simulations.
RESULTS

RESULTS OF THE CASE I SIMULATIONS: For the four different configurations the
simulation and the experimental results were compared. The kinematics of the dummy was evaluated
after every 20ms. For the first configuration of the car motorcycle side impact the results are shown in
the Figure 6.

In the crash test, the dummy helmet just clears the still of the car. In the simulation shown
above, the face guard of the helmet of the dummy impacts the still and the helmet then gets lodged in
the still. This amounts to approximately a 50mm difference in head eight between the simulation and
crash data.

Fig. 5: The complete model of the car-M/C side impact.

Fig 6: Comparison of the simulation kinematics with the crash test at 20msec intervals.

The difference in the results can be assigned to the way the grip of the palm on the handle of
the motorcycle has been modelled. In the simulation no interaction was defined between the palm of
the dummy and the handle of the motorcycle to model the fact that the grip is lost on impact. In the
crash test the hand of the dummy has an encircling grip on the handle. We tested the grip and
measured that it requires a force of about 250N to open the grasp. Consequently, the dummy
continues to hold on to the handle of the M/C well beyond the point when the M/C impacts the car.
This effect is verified in the crash test film where it was clearly seen that the dummy’s hand does not
leave the M/C handle. Without incorporating the complete effect of the M/C dummy hand, we ran
simulations in which the dummy hand stays attached to the handle for 100 ms. The figure 7 shows the
dumps from the simulation, and it is seen that the dummy head does gain height in this case.

Figure 7: Kinematics in the simulation when the dummy hand remains attached to the handle for
about 100ms.

The kinematics of the motorcycle, rider and helmet are reproduced in the simulations. The
point of impact of the head is near the top still, and is very critical as it can end up with the helmet
sliding on the roof or getting lodged on the sill due to a difference of about 30-50 mm in the impact
height. Thus the point of release of the handle has a significant effect on the height gained by the
dummy.

The grip and joints for the MATD seem to be designed to prevent the dummy from collapsing
when seated in the forward leaning posture on the motorcycle. While a better dummy model will
definitely bring the simulations closer to the experiment, a point worth conjecturing is whether the
behavior of the dummy due to the grip does mimic real life situations. For purposes of establishing a
crash standard, these details would influence results significantly.

RESULTS OF THE CASE II SIMULATIONS: In this case the car is moving at 6.7m/s, the
motorcycle is moving at 13.4m/s at an angle of 450 from the front. We see a high sensitivity to the
impact point. As per the [ISO13232] configuration, the front wheel of the motorcycle should impact
the car at the half-length point of the car, which comes out to be 18cm ahead of the B-pillar. The
actual impact location in the test carried out is very close to the B-pillar. Further, through simulations
it is also observed that a slight change in the location of the point of impact causes qualitative change
in the kinematics of the motorcycle and the rider.

The effect of B-pillar is shown in Figure 8. In the first simulation the M/C impacts behind the
B-pillar and the front tire of the motorcycle moves towards the rear side of the car. This matches well
with the experimental results. In the second and the third simulations shown in Figure 8, the front tire
of the motorcycle hits ahead of the B-pillar location. In both cases the tire of the motorcycle moves
towards the front side of the car. The tire seems to be wedged in front of the B-pillar in these cases.
Consequently the car twists the wheel forward. When the impact point is behind the B-Pillar, the tire
slides along the car body as shown in the figure 8.
Fig. 8: Affect of the point of impact of the M/C in the car-M/C Case 2 impact.

In the crash tests it is difficult to control the point of impact to few cm precision in the
presence of lateral movement. Slight error in the experiment may give the results that are not
reproducible. The crash test, in which the impact point was behind the B-Pillar is one such example.

RESULTS OF THE CASE III AND CASE IV SIMULATIONS: For these two configurations
the comparison of the experiment and the simulation is shown in the Figures 9 and 10. There is a
reasonable match and the observations are similar.
Fig 9. Comparison of the simulation kinematics with the crash test at 20msec intervals for the Case III
impact.

In all the cases we have qualitatively discussed the kinematics. As shown, the overall
kinematics depends on many factors. Since many of these factors are not very reproduceable, a
quantitative comparison of the accelerations at this stage is not very meaningful.

Fig 10. Comparison of the simulation kinematics with the crash test at 20msec intervals for the Case
IV impact.

CONCLUSIONS

The qualitative results of the car-M/C side impact simulations presented in this paper show
close match with the experimental observations. The simulations presented highlight the effect of
some of the differences between the H-III and the MATD dummy. These differences play an
important role in the car-M/C impacts. It is also observed that some configurations listed in
[ISO13232] are very sensitive to the impact location. In order to ensure the repeatability of the
[ISO13232] tests it will be important to control the impact location within a small range. Since the
experimental data is limited to four impacts under differing configurations, there is insufficient data to
generate confidence levels in the quantitative data. The repeatability of the crash tests, we feel, also
needs to be established to have a better confidence on future simulations. Also, a Hybrid-III dummy
like model, which deviates somewhat from the MATD, is used in the simulations. Due to the above
uncertainties, a quantitative comparison of accelerations is not relevant at this stage. The limitations
above lead us to fresh areas of study, which are being addressed. These simulations have undoubtedly
thrown new light on many aspects of the car-M/C impact study.

The simulation results presented here match the experimental observations quite closely. It
has been further demonstrated that the affect of some critical parameters in the experiments can be
 demonstrated effectively by altering them suitably in the simulations. We conclude that these
simulations offer an important tool to study the car-M/C (and other) impacts and the sensitivity of
many parameters in them.

REFERENCES

1. [Chawla, 2001] Chawla A, Mukherjee, S, Mohan D, Singh M, Sakurai M and Nakatani T, “A

Methodology for car – motorcycle crash simulation, Jari Research Journal, 2001, Vol 23, No 2, pp
18-21.
2. [Chinn, 1986] Chinn BP, and Macaulay MA, Leg protection of Motorcyclists, Proceedings of the
International IRCOBI conference on the biomechanics of impact, 1986
3. [ISO13232] ISO13232, Motorcycles – test and analysis procedures for research evaluation of
rider crash protective devices fitted to motor cycles, 1996-12.
4. [Nieboer, 1991] Nieboer JJ, Goudswaard AP, Wismans J et al, Computer Simulations of motor
cycle air bag systems, Proceedings of the 13th International Technical Conference on
Experimental Safety of Vehicles, 1991.
5. [Nieboer, 1993] Nieboer JJ, Wismans J, Versmissen ACM et al, Motor cycle crash test modeling,
Proceedings of the 37th Stapp Car crash Conference, 1993.
6. [Takatori, 1999] Takatori O, Deformation Characteristics of tires on frontal collisions and
applications to crash simulations, JSAE proceedings No 10-99, 1999, p17-20.

Back

Back

View publication stats