1
Rear-End Collisions - The Effect of Recliner Stiffness
and Energy Absorption on Occupant Motion
Gert _— Mats Y. Svensson, Per Livsund
Dept. of Injury Prevention, Chalmers University of Technology, Sweden
David C. Viano
General Motors Research & Development Center, Warren, MI, USA.
ABSTRACT
Neck injuries in rear-end collisions constitute a major part of the societal harm associated with road-
traffic. This situation calls for further research in the field. One area of interest is the influence of the seat
on the human body response and interaction during a rear-end impact.
Rear-end collision sled tests and mathematical (MADYMO) simulations were used to investigate the
influence of seat-back stiffness and energy absorption capacity of the seat-back recliner on occupant kine-
matics and biomechanics. The seat used was a front bucket seat with a high head-restraint. In addition to
occupants in normal posture with their back on the seat, occupants leaning forward at the time of impact
were also considered.
‘The results show occupant protection increases with increasing recliner stiffness, up to a threshold
stiffness above which it becomes less sensitive to further changes for rear-end velocity-changes up to 32
mph. The initial position of the head away from the head-restraint was more likely to increase the risk of
neck injury than was impact velocity.
INTRODUCTION
Injury statistics indicate that almost every fourth injury to car occupants is related to rear-end crashes,
and that three quarters of these injuries involve the neck (Data Link 1989; 1990). Neck injuries in rear-
end collisions mostly occur at impact-velocities less than 20 km/h (Kahane, 1982; Romilly et al., 1989;
Olsson et al., 1990), and are mostly classified as AIS 1 (Data Link 1989; 1990; Foret-Bruno et al., 1991;
James et al., 1991; Ono and Kanno, 1993). However, they cause significant harm to society since al-
most ten percent of the AIS I neck injuries occuring in rear-end collisions have been found to lead to
permanent disability (disability-degree 2 10 %) (Nygren, 1984; Nygren et al., 1985).
Rear impacts also involve serious or fatal injury, in addition to the many minor injuries (Data Link
1989; 1990). Such injury is generally associated with large changes in velocity of the struck vehicle,
seat-back deflection, and head impact. In sled-tests with Hybrid-II dummies, Viano (1992) found that
the occupant surrogate was retained on the seat if the seat-back angle with respect to vertical stayed be-
low 60°. When the seat-back deflection exceeded 70°, the occupant was translated rearward, with poten-
tial impact against interior components of the car.
‘The relation between kinematic and kinetic parameters of the head-neck motion and the risk of sustai-
ning neck-injury in a rear-end impact is not fully known. SAE (1993) published limits for neck loads at
the occipital condyles for volunteers and cadavers based on the work by Mertz and Patrick (1967; 1971).
For a volunteer, a bending moment (Y-direction) of 30.5 Nm, a shear force of 231 N (X-direction) and
an axial load of 249 N (Z-direction) was sustained without injury occurring. Melvin and McElhaney
(1972) identified four factors as being important in reducing potential injury in rear impacts: (1) head dis-
placement, rotation, and acceleration; (2) differential motion of the head and torso into the deflected seat-
‘ack; (3) occupant ramping up the deflected seat-back; and (4) occupant rebound.
In 1967 Mertz and Patrick carried out rear-end impact sled tests on a volunteer in a seat with a high,
tigid seat-back. In this study the volunteer's head was in contact with the seat-back during impact. Tests
were done at velocity changes (Av) of up to 30 km/h without the occurrence of injury symptoms. On the
other hand, McConnell et al. (1993) carried out staged rear-end collisions at low impact-velocities in
VI-12
which volunteers were seated in car seats with head-restraints. These volunteers were not exposed to
hyper-extension of the cervical spine during the tests, yet mild and transient, and clinically classical neck
discomfort symptoms were experienced. These results indicate that neck injuries. do not occur during
rear-end impacts if the head is prevented from moving rearward relative to the torso.
States et al. (1970) suggested that the elastic rebound of the seat-back could be an aggravating factor
for the whiplash-extension motion. According to his hypothesis, the rebound of the seat-back can push
the torso forward relative to the vehicle at an early stage of the whiplash extention motion when the head
begins rotating rearward. This in turn would increase the relative linear and angular velocity of the head
telative to the upper torso and could at the same time delay contact between the head and head-restraint,
thus increasing the maximum extension angle. Other studies support this theory (Berton, 1968;
McKenzie and Williams, 1971; Prasad et al., 1975; Romilly et al., 1989; Foret-Bruno et al., 1991;
Svensson et al., 1993a and b). If, on the other hand, the seat-back collapses or yields plastically during a
rear-end collision, the elastic seat-back rebound is eliminated or reduced. In fact, seat-back collapse may
decrease the risk of neck injury in rear-end collisions (Kihlberg, 1969; States et al., 1970; Foret-Bruno et
al., 1991).
Based on field-data and static tests of seat-backs, Warmer et al. (1991) concluded that non-yielding
seat-backs can increase occupant rebound and thus the risk of whiplash-injury. They further concluded
that rigid seats are potentially dangerous to occupants that are not in the normal seated position, i. e. with
their backs away from the seat-back, at impact.
Dummies for rear impact simulations
Because of its human-like shape and mass-distribution, the Hybrid-III dummy has been used as a
human substitute in rear impact tests, (Foret-Bruno et al., 1991; Svensson, et al., 1993a and b; Scott et
al. 1993). While the neck has been validated for extension as well as for flexion (Foster et al., 1977),
the neck attachment at the spine, the stiff thoracic spine, and the short flexible lumbar element and pelvis
may not assure that the Hybrid-III whole-body response is sufficiently human-like for whiplash testing
at low crash speeds. For instance, Scott et al., (1993) showed that the kyfosis of the human thoracic
spine straightens out during rear impact, and this cannot be featured by the dummy. In order to improve
the biofidelity of the Hybrid-III in low-severity rear impacts, Svensson and Livsund (1992) developed
and validated a Rear Impact Dummy-neck (RID-neck) which for rear-impact testing purposes can replace
the standard neck of the Hybrid-III dummy. The reason was that the neck of the Hybrid-III dummy had
been found to generate too high a torque during extension for a human-like response in the sagittal plane
(Seemann et al., 1986 and Foret-Bruno et al., 1991).
Aim
‘The aim of the present study was to investigate the influence of seat-back stiffness and energy absorb-
tion capacity of the seat-back recliner on occupant responses during rear-end collisions of moderate and
high severity. Based on reference sled tests with a Hybrid Il-dummy that was equipped with a RID-neck
and seated in a modern, standard, front bucket seat, mathematical (MADYMO) simulations of the same.
crash situation were conducted. In MADYMO, rear impacts of 12.5 and 32 kmph were simulated with
different stiffness and energy-absorption capacity recliners. The case of the occupant leaning forward
Prior to impact was considered, since occupants may not always travel with their back against the seat.
MATERIALS AND METHODS
Sled tests
In a series of sled tests reported in a previous paper (Nilson et al., 1994), rear impacts with velocity
changes (Av) of 12.5 and 25 kmph (3.5 and 6.9 m/s) were carried out on the sled track at Autoliv AB in
Virgida, Sweden. The crash-pulse had a level of about 70 or 85 m/s*, respectively (Fig. 1).
VE-23
In the tests, a SOth percentile Hybrid [1I-dummy equipped with a RID-neck was used. The dummy
was seated with lower arms resting in the lap. In order to ensure good visibility of the torso motion
during the test, no clothes were put on the dummy upper body. This produced relatively high friction
between the torso and seat-back. A bucket-seat, taken from a midsize car model was used. The dummy
was equipped with accelerometers in the head, chest and pelvis, and with force-moment transducers at
the upper neck (R.A. Denton, type:1716) and at the lower neck (R.A. Denton, type:1794). The sled
acceleration was also measured. The tests were filmed with one high-speed film camera and one high-
speed video camera, both at $00 frames/second.
Mathematical simulations
Model validation
‘The two tests were modelled by MADYMO3D, version 5.0 (TNO Crash Safety Centre, 1992). The static
load-deflection properties of the seat-back and the torque-angle relation of the recliner were measured.
‘The Hybrid-IIl dummy was described according to the database provided by the TNO (TNO Crash
—— 125 ——32 eeee 25
a :
oe \ i
Sled deceleration [m/s?]
>
&
0 0.02 0.04 0.06 0.08 0.1 0.12
Figure 1: The sled pulses used in the simulations. The low severity impact was simulated
by a Av of 12.5 kmph in both the validation and the tests, whereas the high severity impact
was modelled by a Av of 25 kmph in the validation and a Av of 32 kmph in the test with
different recliners.
Safety Centre, 1992), but the single-element neck used in the standard database was replaced by seven
elements, each describing one "vertebra” of the RID-neck (Dusserre, 1993). The moment-angle relation
applied for angular displacement between adjacent vertebrae is shown in Figure 2. The corresponding
data for the Hybrid-III dummy is included for reference. Figure 2 shows neck moment vs. head angular
displacement of the mathematical RID and Hybrid-III necks when exposed to the Hybrid-I1I neck-exten-
sion calibration test. The contact contours of the rearmost parts of the lower torso and the head of the
Hypbrid-IIT dummy were modified in order to better model the interaction of these body regions with the
seat-back and head-rest, respectively. The new contours were based on drawings of the Hybrid-IT
dummy (Fig. 4).
‘The sled-tests were simulated by exposing the dummy and the seat-back to the accelerations of the sled
during the corresponding sled-tests. The coefficient of friction between torso and seat-back (0.6), as well
VI-3