DEVELOPMENT OF A GENERIC LOW SPEED REAR IMPACT PULSE FOR ASSESSING

SOFT TISSUE NECK INJURY RISK

Frank Heitplatz; Raimondo Sferco; Paul A Fay; Joerg Reim; Dieter de Vogel
Ford Motor Company - Köln, Germany and Basildon, UK

ABSTRACT

Five car-to-car full-scale vehicle impact tests were conducted in order to aid the development
of testing protocols for the certification and rating of vehicles, with respect to their ability to protect
occupants from soft tissue neck injures. A wide range of passenger cars (ranging from the Ford Ka
to the Volvo S80) were chosen as the target cars. The Ford Focus, representing the mass of a
typical average sized European car, was used as the bullet car. All the tests were 100% overlap
rear impacts which aimed to achieve a change in velocity (or “∆V”) of 16 km/h for the target
vehicles. Due to the different ride heights of the target vehicles, the tests included over- and under-
ride impacts as well as bumper-to-bumper impacts. The results showed significant differences with
respect to signal shape and peak acceleration levels when compared with existing proposals. In
particular, the crash pulses showed a bi-modal nature and had a high initial peak. These
characteristics appear to be typical for current production passenger cars. The recorded crash
pulses have been numerically combined into a proposed single pulse. Finally, a tolerance corridor
for the sled acceleration signal and the sled ∆V has been proposed.

KEYWORDS
• WHIPLASH
• ACCELERATIONS
• CAR TO CAR IMPACT TESTING
• REAR IMPACTS
• REGULATIONS

SOFT TISSUE NECK INJURIES (STNI) are one of the most frequent types of injury seen in
motor vehicle crashes. Recent studies have reported that these injuries (Temming et al. 1998, Hell et
al. 1999, Krafft et al. 2000, 2001, Linder et al. 2001):
• Occur frequently (but not exclusively) in rear impacts
• Occur at relatively low speeds
• Appear to be reported more frequently than they did 10 or 20 years ago
• Carry a large burden for individuals, society in general and the insurance industry in
particular.
There is therefore considerable interest in reducing the likelihood of STNI, particularly from the
insurance industry who would like to reduce the level of pay outs for claims of neck injury.
A lot of research effort is currently focussed on the development of suitable test methods for
assessing STNI risk (Hell et al 1999, Anselm et al. 2001). At present however, there is not a full

1
understanding of the mechanisms of STNI and there is no universally consensed view about the choice
of test dummy, injury criteria, and test mode or test severity. A suitable test for assessing STNI could
potentially be used for consumer safety testing (for example, by Euro-NCAP) (Klanner 2001), for
determining insurance ratings (Avery 2001) and for setting minimum protection levels in legislation
(Langwieder 2000). There are clear benefits if there is a harmonised test procedure which could be
used for each of these three different purposes. Apart from the savings in test costs and the ability to
share data, it is also easier for manufacturers to produce optimised designs if they have one set of
targets rather than numerous, potentially conflicting, requirements. A harmonised test method would
need to be realistic, repeatable and reproducible.
The study described in this paper attempts to address the open issues concerning the appropriate
test severity for assessing STNI risk in modern passenger cars in rear impacts. Here there is a choice
between carrying out complete vehicle impact tests or replicating such impacts using a sled test. In
general, sled tests (using a “seating buck” on a sled rather than a complete vehicle or bodyshell) are
preferred over complete vehicle tests (Langwieder et al. 2000). They are cheaper and quicker to carry
out and it is usually easier to see and analyse what is happening to the dummy’s neck. From a
legislative perspective, a generic sled test makes it possible to approve a seat design for different
models without the need for numerous tests and it is also possible to approve a new seat design
without the need for destructive testing of a complete vehicle.
A potential difficulty in performing sled tests however is determining the appropriate crash pulse.
In a real vehicle impact, an occupant will experience a unique crash pulse based on the structural
characteristics of both his own vehicle and the impacting vehicle, the seat and head restraint
characteristics, the vehicle speeds involved, influence of braking and road surface friction, etc. There
is therefore no single “right” answer to the question of what crash pulse to use in a sled test. Instead, it
is proposed that a generic pulse (based on the response of typical modern vehicles in car-to-car tests)
could be the most appropriate approach to follow.

METHODOLOGY

Five full-scale car-to-car crash tests were conducted at Ford's Crash Laboratory at Köln in
Germany. The tests were 180° rear impacts with 100% overlap. The number of tests that can be
conducted is always limited by the very high cost for this kind of test. Therefore, not all possible
impact scenarios i.e. heavy braking of the bullet vehicle or variations in bullet vehicle size could be
investigated. However, the test series was able to investigate the effect of varying weights of the
impacted vehicle. The target vehicles ranged from the smallest currently produced Ford car (the Ford
Ka) to one of the largest passenger cars produced within the Ford Motor Company brands (the Volvo
S80). The matrix of tests is shown in Figure 1.
Bullet Vehicle Test-weight (kg) Target Vehicle Test-weight (kg) Bullet impact speed (km/h) Target delta V (km/h)
Ford Focus 4dr 1300 Ford Ka 1021 26.2 16.9
Ford Focus 4dr 1301 Ford Fiesta 3dr 1165 26.5 16.2
Ford Focus 4dr 1303 Ford Focus 5dr 1384 26.5 13.7
Ford Focus 4dr 1299 Ford Mondeo 5dr 1675 32.7 15.7
Ford Focus 4dr 1301 Volvo S80 4dr 1860 33.9 16.9
Figure 1: Test matrix
The bullet vehicles were identical Ford Focus 4 door saloons. The test weight of 1300 kg was
chosen to represent the typical weight of an averaged sized European car.
All bullet and target vehicles were equipped with accelerometers at the B-pillar to doorsill joints.
They were loaded with 95% of their fuel tank capacity using a test fuel substitute and had two 50%ile
male dummies placed in the front seats to get a realistic on-road weight distribution. In order to
achieve the final bullet car test weight of 1300 kg, components were removed or weights rigidly
attached as appropriate. The ride-height of the bullet vehicles was left “as found” and not modified in
any artificial way.

2
 On the target vehicles, test instrumentation was rigidly attached to the vehicle bonnet. The weight
of the instrumentation was partly offset by removal of components in the engine compartment and
drainage of liquids from the engine. However, the instrumentation did lead to some forward pitch of
the vehicles but this did not significantly affect the actual ride height of the rear bumper system. An
example test set-up is shown in Figure 2.

Figure 2 Test Set-up Example: Focus to Mondeo

The tests were set up with the aim of achieving a ∆V of 16 km/h for the target vehicle. This
choice of severity was based on an analysis of German insurance data which concluded that the speed
range between 10 and 20 km/h ∆V represents the main problem area. However, the lower speed range
(∆V<15 km/h) was associated with a large number (50%) of cases which had insufficient or
implausible medical documentation (Langwieder et al. 2001). It has also been proposed by ISO (ISO
1998) and the International Insurance Whiplash Prevention Group (IIWPG) (Avery 2001) to use a ∆V
of 16 km/h in their test method development. The impact speed of the bullet vehicles in this study was
therefore adjusted to give a ∆V of 16 km/h for the target vehicle depending on the relative weight of
the target and bullet vehicles. Some degree of iteration of the impact speed was necessary in order to
achieve the desired ∆V of 16 km/h. For example, the first test conducted (Focus to Focus) actually
resulted in a ∆V of only 13.7 km/h. All following tests did fall into the tolerance band of ±1 km/h to
the ∆V of 16 km/h
In order to allow a detailed analysis of the structural bumper interaction on high-speed film, the
plastic bumper skin was removed from those vehicles where the plastic skin was in direct contact with
the underlying steel bumper. It was assumed that in these cases the cosmetic components have no
significant effect on the interaction between the vehicles.

RESULTS

Impact Mode - Example shots from the high-speed film of some impacts are shown in the
Appendix. Close examination of the film allowed the mode of vehicle interaction to be determined.
In particular, it was possible to evaluate if the impact was an over ride, bumper-to-bumper impact or
an under ride. These results are shown in Figure 3 (see also Figure 8 to Figure 9 in the Appendix).

3
 Test Impact class
Focus to Ka Under ride
Focus to Fiesta Over ride
Focus to Focus Under ride
Focus to Mondeo Bumper to Bumper
Focus to S80 Bumper to Bumper
Figure 3 Impact classes
Crash Pulses – The crash pulses from the individual tests are given in Figure 10 to Figure 14 in
the Appendix. All the LHS/RHS average acceleration traces (filtered at CFC60 according to
SAEJ211) reached their peak in the first 25 msec of the impact. At this time in the crash, the
structures of the vehicle were in full engagement and the energy-absorbing crush of the bumper was
complete. Comparing acceleration levels after 10 msec for the bumper-to-bumper interactions (eg
Mondeo 6g) and with the over-rides (eg Fiesta 4g) or the under-rides (eg. Ka 2g) it was found that the
initial acceleration is higher for direct bumper interactions. There is a period of low acceleration of
the target when the bullet under rides. This effect led to a significant shift in time (Approx. 10 ms).
However, eventually the structures were fully engaged with each other leading to similarly high peaks
as in bumper to bumper impacts. The level of the peak acceleration seemed to be relatively
independent of under/override or direct bumper-to-bumper engagement. There was no apparent
correlation between vehicle weight/class and the levels of peak or mean acceleration in this test series
(See Figure 4).
Impact Test weight Acceleration peak CFC60 Time of peak Mean acceleration
(LHS/RHS Average)
Focus – Ka 1021 17 25 4.3
Focus – Fiesta 1165 14 28 4.2
Focus – Focus 1384 10 17 5.0
Focus – Mondeo 1675 15 16 4.7
Focus – S80 1860 9 20 4.6
Figure 4 Peak and Mean Acceleration Levels (The mean is calculated between t = 0 msec and the point at which
the trace falls below 1g at the end of the impact event).

Generic Pulse - Due to the time shift in over/under rides, simple overlaying and averaging of the
recorded CFC60 pulses give a misleading picture as the peaks of the under-ride pulses correspond to
valleys in the bumper to bumper pulses (see Figure 15). In order to generate a generic pulse corridor
signals where shifted in time so that all the signals had their onset of crash significant acceleration at
t=0 msec (see Figure 5). Displaying the signals in this manner shows a clear quantitative and
qualitative correlation. There is a significant initial rise in acceleration, peaking at a time when the
bumper crush is complete. This is followed by a characteristic “bi-modal wave” in the acceleration
traces.

4
 Average of all Pulses Compensated for Shift in Onset of Acceleration

160
150 Fiesta
140
130
Ka
120 Average pulse CFC60
110 with time shift
100
90 Focus
acc (m/s^2)

80
70
60
50
40 S80
30
20
10
Mondeo
0
-10 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0.11 0.12
-20
-30
time (s)

Figure 5 Acceleration traces from test series filtered with CFC60, shifted to common t=0.

A separate CAE investigation (not reported in detail in this paper) has shown that this
characteristic shape of the acceleration traces is caused by the back-swing of the free engine mass
(causing a trough in the acceleration trace) followed by a new onset of acceleration as structural
deformation resumes. All current mass-produced passenger vehicles share the same basic design
features (eg reinforced rear structures and front mounted free-swinging engines). It is therefore
reasonable to assume that for this test set-up the general features of the bi-modal pulse would also be
exhibited by vehicles not included in this study.
In order to define a generic pulse corridor, an average of all the filtered and time-shifted signals
was calculated. For simplification the very low signal levels after 90 msec were ignored. A
acceleration corridor in which the sled pulse should fall were generated based on averaged pulse. In
addition to these corridors, it is also proposed that a tolerance of ±1km/h should be applied to the
proposed ∆V of 16km/h. The proposed generic pulse and its associated corridors are shown in Figure
6. The same figure also shows some of the other existing or proposed pulses.

5
 Generic Pulse and Corridor

13.0

12.0 New proposed upper limit GDV proposed lower limit

11.0 Generic pulse (CFC60 including GDV proposed upper limit

9.0

8.0
Acceleration (g)

7.0

6.0

5.0

4.0

3.0
Thatcham proposed lower limit
2.0 (no upper limit)

1.0

0.0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Time (s)

Figure 6: Comparison of currently proposed sled pulses

DISCUSSION
The results from the tests reported in this study have been used to propose a generic pulse for
dynamic rear impact sled testing. The proposed pulse is characterised by a fairly high initial peak
acceleration and a bi-modal wave form in the acceleration trace. It is believed that this pulse is
appropriate to most modern car designs. Although the data sample for this study is comparatively
small, the results are in line with data published for vehicles from other manufactures (Avery 2001).
A simulation of a HyGe sled test has been undertaken to determine if the proposed pulse can actually
be reproduced on a commercial HyGe sled. The simulation clearly showed that the proposed pulse
corridor is feasible for HyGe sled testing (see Figure 7).

HyGe simulation
Result

Theoretical best fit to the

Figure 7: HyGe simulation of the generic pulse

6
 Other generic pulses have been proposed by GdV and Thatcham. A comparison of these pulses
with the new proposed generic pulse is shown in Figure 6.
The GdV pulse was based on analysis of data from older vehicle models, designed prior to the
mid 1990's (Langwieder et al 2000). This proposal does not show a distinct initial acceleration peak
but instead has a slow rise in acceleration up to a plateau level of 6.5g. Most of the crash pulse data
used to generate this proposal came from vehicles manufactured in the 1980's and it is felt that vehicle
design has changed significantly enough to warrant a re-assessment based on more recent vehicle
designs.
The Thatcham proposal (Avery 2001) is based on more modern vehicles and goes some way
towards replicating the fall of in acceleration seen after the initial peak. However, due to the
prescribed lower limit corridor (an upper limit is not given) that produces an ∆V of 15.7 km/h, it is
practically not possible to have initial peaks of more than 8g if the resulting ∆V is not to be larger than
16 km/h. However, the results reported in this study show peak accelerations of between 9g and 17g
(LHS/RHS average, filtered with CFC60). As a result of this study, it is felt that a generic pulse ought
to specify initial acceleration peaks of around 10g, with corridors allowing peaks up to 11.5g.

CONCLUSIONS
1. Five car-to-car full-scale vehicle impact tests were conducted in order to aid the development
of testing protocols for the certification and rating of vehicles, with respect to their ability to
protect occupants from soft tissue neck injures.
2. The results show significant differences with respect to signal shape and peak acceleration
levels when compared with existing proposals. In particular, the crash pulses showed a bi-
modal nature and had a high initial peak. These characteristics appear to be typical for current
production passenger cars.
3. The recorded crash pulses have been numerically combined into a proposed single pulse
suitable for sled testing and a tolerance corridor for the sled acceleration signal and the sled
∆V has been proposed.

ACKNOWLEDGEMENTS
The authors would like to thank Volvo Cars for their help and support during the test program.

REFERENCES

Anselm D, Deutscher C, Grossmann H, Protecting Car Passengers in Rear-End Collisions, Proc. of

Avery M, Car to car testing and crash pulse selection, Proc. of IIWPG/IRCOBI Symposium on Dynamic Testing
for Whiplash Injury Risk, Isle of Man, UK, 2001

Hell W., Langwieder K. Consequences for seat design due to rear end accident analysis, sled tests and possible
test criteria for reducing cervical spine injuries after rear-end collision. Proc. of the Int. IRCOBI Conf. on the
Biomechanics of Impacts, Sitges, Spain, 1999.

ISO Road Vehicles – Traffic Safety Analysis – Part 2: Determination of Impact Severity ISO/TC22/SC12/WG7,
IS/WD 12353-2. Draft Version 1998

Klanner W, Status Report and Future Development of the Euro NCAP Programme, Proc. of 17th ESV
Conference, Amsterdam, Netherlands, June 2001.

Krafft M, Kullgren A, Ydenius A, Tingvall C, The Correlation between Crash Pulse Characteristics and Duration
of Symptoms to the neck – Crash Recording in Real Life Rear Impacts, Proc. of 17th ESV Conference,
Amsterdam, Netherlands, June 2001.

Krafft M, Kullgren A, Tingvall C, Boström O, Fredriksson R. How crash severity in rear impacts influences

7
 short- and long-term consequences to the neck. Accident Analysis and Prevention 32 (2000) pp187-195.

Langwieder K, Hell W, Proposal for an international harmonized dynamic test standard for seats and head
restraints, Proc. of IIWPG/IRCOBI Symposium on Dynamic Testing for Whiplash Injury Risk, Isle of Man,
UK, 2001

Langwieder K., Hell W., Schick S., Muser M., Walz F and Zellmer H. Evolution of a dynamic seat test standard
proposal for a better protection after rear-end impact. IRCOBI Conference – Montpellier France, 2000.

Linder A, Avery M, Krafft M, Kullgren A, Svensson M, Acceleration pulses and crash severity in low velocity
rear impacts-real world data and barrier tests, Proc. of 17th ESV Conference, Amsterdam, Netherlands, June
2001.

Temming J, Zobel R, Frequency and Risk of Cervical Spine Distortion Injuries in Passenger Car Accidents:
Significance of Human Factors Data. Proc. of IRCOBI Conference on Biomechanics of Impacts, Gotheborg,
Sweden, 1998: pp. 219-233.

8
APPENDIX – HIGH SPEED CAMERA SHOTS AND ACCELERATION TRACES FROM CAR-
TO-CAR TESTS

0 msec 80 msec
Figure 8 Focus to Ka

0 msec 80 msec
Figure 9 Focus to S80

9
 Focus to Ka

180

160 LHS_B_POST_BASE_LONG
RHS_B_POST_BASE_LONG
140

120

100
acc (m/s^2)

80

60

40

20

0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0.11 0.12
-20
Time (s)

Figure 10 Pulse diagram Ka

Focus to Fiesta

200
180
LHS_B_POST_BASE_LONG
160 RHS_B_POST_BASE_LONG
140
120
100
80
acc (m/s^2)

60
40
20
0
-20 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0.11 0.12
-40
-60
-80
-100
Time (s)

Figure 11 Pulse diagram Fiesta

10
 Focus to Focus

180
LHS_B_POST_BASE_LONG
160
RHS_B_POST_BASE_LONG
140

120
acc (m/s^2)

100

80

60

40

20

0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0.11 0.12
-20
Time (s)

Figure 12 Pulse diagram Focus

Focus to Mondeo

180
160 LHS_B_POST_BASE_LONG
140 RHS_B_POST_BASE_LONG

120
100
80
acc (m/s^2)

60
40
20
0
-20 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0.11 0.12

-40
-60
-80
Time (s)

Figure 13 Pulse diagram Mondeo

11
 Focus to S80

180

160 LHS_B_POST_BASE_LONG
RHS_B_POST_BASE_LONG
140

120
acc (m/s^2)

100

80

60

40

20

0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0.11 0.12
-20
Time (s)

Figure 14 Pulse diagram S80

Average of all Pulses Without Time Shift

160
150
140
130
120 Average pulse CFC60
110 without time shift
100
90
acc (m/s^2)

80
70
60
50
40
30
20
10
0
-10 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0.11 0.12
-20
-30
time (s)

Figure 15 Average of all pulses without compensation for over-under ride time shift

12