THE INTERACTION OF AIR BAGS WITH UPPER EXTREMITY TEST DEVICES
C.R. Bass
S.M. Duma
J.R. Crandall
S. George
University of Virginia
S. Kuppa
Conrad Technologies Inc.
N. Khaewpong
E. Sun
R. Eppinger
National Highway Traffic Safety Administration
United States
Paper Number 98-S7-O- 12
ABSTRACT
This study examines and compares the response of
two upper extremity test devices under driver-side air bag
deployment to contribute to the development of dummy
surrogates for the investigation of primary contact
forearm injuries during air bag deployments. The first of
these test devices, the SAE 5’ Percentile Female Arm
(SAE Arm), is an anthropomorphic representation of a
small female forearm and upper arm that is instrumented
with load cells, accelerometers and potentiometers to
enable the determination of upper extremity kinematics
and dynamics. The second, the Research Arm Injury
Device (RAID), is a simple beam test device designed for
detailed investigation of moments and accelerations
resulting from close contact in the initial stages of air bag
deployment. The RAID includes strain gauges distributed
along its length to measure the distribution of moment
applied by the air bag deployment.
The study used four air bags representing a wide
range of aggressivities in the current automobile fleet.
The upper extremity position was a ‘natural’ driving
posture when turning left with one hand across the
steering wheel. The forearm was positioned directly on
the air bag module with the forearm oriented
perpendicular to the air bag module tear seam. For the
SAE Arm, the humerus was oriented normal to the
steering wheel. Tests with the SAE Arm were performed
both with the arm attached to a 5’ Percentile Female
Hybrid III dummy and with the arm mounted to a
universal joint test fixture. TheRAID was mounted to an
articulated test fixture. In addition to the dynamic tests, a
detailed comparison of the inertial properties of each of
the test devices with the inertial properties of a typical
small female was performed.
Forearm response from both test devices confirmed
the levels of air bag aggressivity determined using
previous cadaveric injury results. In addition, logistic risk
1628
functions for forearm fracture were developed using
existing cadaver studies and the moment response of each
test device. These risk functions indicate that for 50%
risk of ulna or ulna/radius fractures, the SAE arm peak
forearm moment is 61 N-m (+/- 13 N-m standard
deviation) while the RAID peak forearm moment is 373
N-m (+/- 83 N-m standard deviation). For 50% risk of
fracture of both the ulna and the radius, the SAE arm
peak dummy forearm moment is 91 N-m (+/- 14 N-m
standard deviation) while the RAID peak forearm
moment is 473 N-m (+/- 60 N-m standard deviation).
INTRODUCTION
Although the use of air bag systems as supplemental
restraintsh as significantly decreasedth e risk of fatality in
automobile collisions, there is evidence of increased risk
of non-fatal injuries including burns, abrasions, and eye
injuries owing to air bag deployment. In addition, case
studies suggest that upper extremity injuries, including
severe fractures, may be caused by air bag deployment
[c.f. Marco 1996, Freedman 1995, Huelke 1995,
Kirchoff 1995, and Roth 19931. Kuppa et. al. analyzed
several accident databasesto determine the incidence of
upper extremity injury for accidents with and without a
driver-side air bag deployment [Kuppa 19971. They
found that 1.1% of drivers who were restrained only by a
seatbelt experienced an upper extremity injury. In
contrast, 4.4% of drivers experienced upper extremity
injuries in the presence of a deploying air bag.
Two modeso f injury have been suggestedto explain
this increased incidence of upper extremity injuries with
air bag deployment. The first type is a flinging type of
injury in which the air bag propels the arm into an object
in the vehicle (e.g. b-pillar, roof, and occupant’s head).
The second type is primary contact with the air bag or air
bag flap; this injury may occur, for example, while
executing a left turn with a continuous motion of the right
hand, placing the forearm directly over the module. It is
the latter group, primary contact injuries, that is the
subject of the current study.
Case studies and NASS data suggest that these
severe upper extremity injuries occur predominantly in
women. It may be hypothesized that this represents the
effects of three factors: 1) as women are generally shorter
in stature than men, they drive closer to the steering
wheel/air bag module, 2) women experience an agerelated
loss of bone mineral density, and 3) women have
generally smaller bones and, hence, lower ultimate bone
strength.
To investigate the upper extremity/air bag
interactions causing these injuries, Saul et al used an
instrumented 50’i’ Percentile Male Hybrid III upper
extremity to examine injury from direct contact [Saul
19961. Using strain gauges and accelerometers, they
found that bending moments and accelerations of the
forearm could be accurately recorded. Moreover, a
correlation was found between these values and the air
bag’s inflator properties, flap, and steering wheel
orientation. In addition, the forearm bending moment
response of the instrumented SAE 5’h Percentile Female
Arm (SAE arm) under primary air bag contact has been
correlated with cadaveric injury to produce an injury risk
function for small females [Bass 19971.
In addition, the Research Arm Injury Device
(RAID) was developed by Conrad Technologies Inc. and
NHTSA to investigate the interaction between a
deploying air bag and an upper extremity in close
proximity to the air bag [Kuppa 19971. They found that
the two most significant determinants of peak measured
bending moment were the orientation of the arm with
respect to the air bag module and the separation distance
between the two. Maximum moments were recorded
when the forearm was positioned perpendicular to the air
bag module. This situation occurs, for example, when
making a left turn with the right hand. In,this situation,
the right and left sides of the air bag are at the 1 and 7
o’clock positions respectively, while the hand and elbow
are at the 10 and 4 o’clock positions respectively. The
maximum moments also decreased as the distance
between the air bag and the forearm was increased from
1.3 cm to 7.6 cm.
It is likely that a specific air bag design is developed
with a view toward total restraint system effectiveness.
As different passenger automobiles have different
physical sizes and stiffness, this results in installed air
bags of different deployment properties (e.g. pressuretime
histories, module design, and deployment
characteristics) among vehicle models. Four OEM air
bag types were used in this study; these air bags were
identified using RAID testing as representing a wide
range of aggressivities in the current passenger car fleet.
Using a previously created coding scheme [Bass 19971,
these systems are termed System H, System K, System J,
and System L air bags. The System H and System K air
bags produce relatively more aggressive air bag
deployments, the System J air bag produces a moderately
aggressive deployment, and the System L air bag
produces a relatively less aggressive deployment. In
addition, the System H air bag has been identified in case
studies as producing primary contact upper extremity
injuries under certain circumstances.
The principal goal of this study is to examine the
suitability of both the SAE arm and the RAID in
characterizing the forearm forcing during air bag primary
contact using OEM air bag systems. In addition, this
study quantifies the dynamic response of dummy upper
extremities under air bag deployment in a ‘worst-case’
position. Also, the study investigates factors that affect
injuries in cadaveric upper extremities and develops a
correlation of these injuries with dummy response using
the SAE Arm and the RAID. As there are a number of
design factors that may influence the upper extremity
injury potential of a given air bag, including inflator
properties, air bag properties and module properties, we
have chosen to focus on dummy and cadaveric response
criteria as the most effective measure of injury risk.
The testing was performed in two major parts. The
first includes tests of the RAID test device under air bag
deployment in a representative ‘worst-case’ position for
air bag deployment. The second is a study of the same set
of deploying air bags into the SAE arm mounted on a
Hybrid III dummy and tests with the SAE arm attached to
a universal joint arm fixture developed for cadaveric
studies [Bass 19971. This second series of tests involves
forearm positioning similar to that prescribed for the
RAID testing.
TEST DEVICES
Several instrumented dummy arms exist that are
appropriate for use in arm/air bag interaction studies;
these include the 50% Male Hybrid III Instrumented Arm
[Saul 1996, Johnston 19971, the Research Arm Injury
Device (RAID) [Kuppa 19971, and the SAE 5fi Percentile
Female Instrumented Arm (SAE arm) [Bass 19971. As
the epidemiogical analysis of air bag-induced upper
extremity injuries suggests that small females suffer
injuries at a much greater rate than males, this study
investigates the use the SAE arm and the RAID as
suitable dummy surrogates for the development of risk
functions using previously reported small female
cadaveric injury studies [Bass 19971.
A diagram of the SAE arm is shown in Figure 1.
Pronation/supination of the forearm is provided by a
1629
-
single degree-of-freedoma xial 360’ rotation in the wrist.
The forearm is a single shaft incorporating a six-axis load
cell located approximately mid-shaft. The elbow is a
single degree-of-freedom clevis joint allowing elbow
flexion/extension with a soft joint stop in each direction.
This elbow motion may be measured using a
potentiometer incorporated into the elbow. In addition,
strain gauges to measure two bending axes are located in
the distal humerus. The humerus is a single shaft with a
six-axis load cell approximately midshaft. At the
proximal end of the humerus, two degree-of-freedom
rotations are allowed by a 360’ axial rotation at the top of
the humerus shaft and a clevis joint at the shoulder. In
addition to the existing instrumentation on the SAE arm,
the current study addeda distal triaxial accelerometear nd
a single-axis MHD angular rate sensor mount located one
third of the distance from the wrist to the elbow.
Additional accelerometemr ounting locationsi n the elbow
were not used.
Motions allowed by the SAE arm listed in Figure 2
are approximately anthropomorphic with the exception of
pronationfsupination and shoulder motions. For
pronationlsupination, the existence of a single shaft
forearm limits both the availability and the utility of
forearm rotations located outside the wrist. Though the
predominant flexion/extension motions and upper
humerus rotations are represented in the SAE dummy
shoulder, the human shoulder has three degrees-offreedom
in rotation and limited translation that is not seen
in the dummy.
In contrast, the RAID, shown in Figure 3, has a
more limited range of motions. Developed as an
investigative tool to study primary contact arm/air bag
interactions, the RAID is constructed of a 3.2 mm thick
aluminum tube of 5 1 mm diameter with a two degree-offreedom
clevis joint to allow rotational motion along two
axes. The mass of the tube (1.6 kg) was chosen to
approximate a 50th percentile male human forearm. To
simulate the effects of a hand, a small additional mass
(0.5 kg) is attached to the free end of the RAID. The
length of the RAID was selected as 460 mm to protect the
pivot attachmentsf rom the deploying air bag. The RAID
instrumentation includes five stations of diametrically
opposed strain gages to measure moments along two
axes. In addition, rotations are measured by two angular
potentiometers,a nd triaxial accelerationsa re measureda t
the approximate mid-length of the RAID. The RAID is
covered with 20 mm of foam and rubber skin similar to
that on the Hybrid III mid-forearm.
As the RAID incorporates simple two-dimensional
rotation, the RAID simulates only the forearm degrees of
freedom associated with elbow flexiotiextension and
shoulder abductionladduction. So, while the RAID may
be appropriate for primary contact with a deploying air
bag, it is likely not appropriate for later interactions
involving additional upper extremity degrees of freedom.
Angular Rate Sensor Triax Elbow Joint
Accelerometer Triax
7
Strain Gauges
7
i
Elbow Joint i
Rotary Potentiometer”^‘W~W-~’
Angular Rate Sensor Triax
Accelerometer Triax
Figure 1. Picture of the SAE 5th Percentile Female Instrumented Arm (SAE Arm).
1630
Axial Rotation,
Flexion/Extension,
AbductionlAdduction )y’-“‘-\,
Axial Rotation
Flexion/Extension
AbductionlAdduction FlexionlExtension
Figure 2. Motions of the SAE 5th Percentile Female Arm.
U&4Xial
(4 at each station)
Accelerometer
/-
/ Mounting Block (3 Accelerometers)
educed Diameter
/Rotational
tentiometers
‘\__ HAND MASS
Figure 3. Research Arm Injury Device (RAID).
A comparison of the segment masses of the SAE
arm and the RAID with the 5” and 50” percentile female
population are shown in Error! Reference source not
found.. The SAE arm is substantially heavier than the
reference 5” percentile female population but is similar to
the reference 50ti percentile female population. The
RAID, however, was designed to simulate a 50”
1631
percentile male. So the RAID is substantially heavier
than the forearms of the reference female populations.
Reference forearm and hand lengths shown in
Error! Reference source not found. were derived from
an anthropometric study on 1905 USAF women
[McConville 19791. For the human population, the
forearm length is taken to be the distance from the tip of
the olecranon to the tip of the ulna styloid process, and
the hand length is the distance from the ulna styloid
process to the middle finger of the outstretched hand.
The dummy arm measurements are taken from the
rotation centers.
The total forearm/hand length of the SAE arm of
405 mm is comparable to the 5* percentile female length
of 397 mm but over 20 mm less than the forearm/hand
length of the reference 50ti percentile female. In contrast,
the total length of the RAID (460 mm) is much larger
than the forearm/hand length of the reference female
population and is comparable to the forearm/hand length
of a 50* percentile male population (491 mm) but
significantly larger than the forearm length of a 50”
percentile male population (299 mm). So, the SAE
dummy forearm is similar to a 5’h percentile female
population in length but a 50th percentile female
population in mass ,while the RAID is, by design, similar
to the 50th percentile male in mass and forearm/hand
length.
Table I. Comoarison of Reference and Dummv Arm Anthrouometrv
Three-wire torsional pendulum studies were
performed on the segments of the SAE arm to determine
inertial properties in the principal axes. Axes of rotation
passing through the segment center of gravity define all
moments, and the reference female forearms are oriented
in the neutral position. The x and y principle axes of the
dummy and reference female forearms are approximately
normal to the anatomical axis running along the forearm.
The z principle axis is approximately tangential to this
axial axis. Moments of inertia were calculated for+ the
RAID assuming a uniform aluminum cylinder.
For primary contact injury under air bag
deployment, kinematics observed in previous dummy and
cadaver studies [c.f. Bass 19971 indicates that there is no
significant motion of the humerus prior to peak moments
or cadaveric injury. So, the inertial properties of the
upper arm are negligible in the investigation of surrogate
response under primary contact air bag deployment.
Also, the dynamic significance of pronatiotisupination
(axial rotation of the forearm) motions is minimal in
primary contact studies, so the principle moment of
inertia about the axial axis is of limited significance in
this study.
For the SAE arm, the influence of the mass of the
centrally located load cell on the x and y principle
moments of inertia is clear. Though significantly heavier
than the reference 5’h percentile female reference
population, the SAE arm has x and y moments of inertia
that are comparable to the reference 5’ percentile female
population. A significant portion of the mass of the SAE
forearm is included in this load cell. The z (axial)
moment of inertia of the SAE arm, however, is larger than
the 5’h percentile female owing to the size of the SAE
arm. In addition, owing to the substantial mass of the
SAE hand, principle moments of inertia in the x and y
axes are much larger than those of the reference 5’h
percentile female population, and are more comparable to
those of the 50” percentile female.
For the RAID, the length is significantly greater than
the forearm of either female reference population or that
of a 50* percentile male. So, the moment of inertia of the
RAID forearm segment is substantially larger than that of
either reference female population. In air bag tests, this
’ [McConville 19791
* RAID is single segment.
1632
moment of inertia (z axis) of 1040 kg-mm2 is
Table 2. Principle Moments of Inertia
will likely result in lower peak velocities and possibly commensurate with a 50* percentile male value of 1180
much larger moments. This imposes an additional kg-mm2 [McConville 19791. The ‘hand’ mass of the
limitation on the use of the RAID in the investigation of RAID can be considered to be concentrated at the end of
‘flinging’ injuries in which maximum velocity plays an the RAID for the purpose of this study as the test device
important role in injury mechanics. The RAID axial allows only rotations about the other end.
EXPERIMENTAL SETUP
Both the RAID and the SAE arm attempted to attain
a ‘worst-case’ test condition and hence a ‘worst-case’
response under air bag deployment. The test position
selected is roughly a ‘natural’ driving position in a onearmed
left turn maneuver modified for enhanced
repeatability and ‘worst-case’ behavior. The SAE
forearm was placed directly on the air bag module with
the forearm oriented perpendicular to the air bag tear
seam as shown in Figure 4. The distal third of the SAE
forearm was placed over the module tear seam, and the
humerus was oriented normal with respect to the plane of
the steering wheel. In this configuration, the dummy
fingers do not reach the steering wheel for any of the
OEM air bags tested. Positioning was maintained using
frangible tape.
This position represents the ‘worst case’ or most
vulnerable position for four reasons. First, previous
RAID testing indicated that bending moments were
maximized when the test device was oriented
perpendicular to the air bag tear seam [Kuppa 19971.
Second, RAID bending moments under air bag
deployment were found to decrease as the test device was
moved away from the module. Though the RAID was
placed at distances 13 mm and greater from the air bag
module, out-of-position thoracic testing [Melvin 1993,
Bass 19981 suggests that positioning directly on the air
3 [McConville 19791
bag module may constitute a worst case for certain
occupant/air bag interactions. Third, the distal third of
the human forearm is the weakest location in bending
with the lowest combined polar moment of inertia of both
the radius and ulna, providing the greatest risk of fracture.
Fourth, the humerus oriented normal to the steering wheel
provides a support for the forearm under air bag
deployment forcing the initial center of forearm rotation
to be about the elbow with a relatively long moment arm.
hand
Config.
V
Figure 4. Test Configuration - Arm Relative to Steering
Wheel,
Eight of ten SAE arm tests were performed on a
universal joint test fixture diagrammed in Figure 5. The
fixture is comprised of two components. The first
supports the steering wheel/air bag module on a five-axis
load cell. The second mounts the arm to a four degree-of-
1633
freedom universal joint. A five-axis humerus load cell
was mounted at the interface between the SAE arm and
the universal joint at the shoulder. For the fixture tests,
the center of rotation of the universal joint was located at
a position equivalent to the center of rotation of the
Hybrid III shoulder joint relative to the humerus. The
remaining two tests were performed with the SAE arm
attached to the Hybrid III 5’ percentile female dummy.
One possible objection to the use of the test fixture
is that, for experimental convenience, the location of the
point about which the shoulder rotates is fixed in space.
In a natural driving condition, the shoulder is relatively
free to translate in response to forcing. This
translationally fixed shoulder was examined using the
Articulated Total Body (ATB) lumped-mass simulation
program as shown in Figure 6. The figure shows a
comparison of the humerus axial force for a subject with a
shoulder fixed in translation versus a shoulder free to
translate under the action of a deploying air bag. There is
little difference in humerus response between the two
cases, especially in the crucial initial deployment period.
This result justifies using a shoulder that is fixed in
translation for the experimental setup.
z
Padding
I X -I
Airbae module I
Load cell Universal Joint
I 1 &eering Wheel
1 1 Fixture
Figure 5. Arm/Air Bug Test Fixture.
Figure 6. ATB Simulation of Fixed vs. Sliding Shoulder.
A side view of the test setup with the RAID is
shown in Figure 7. The RAID hangs vertically in front of
the steering wheel and rotates at the mounting pivots.
The test device may be translated in three dimensions to
achieve desired positioning with respect to the air bag
module. For this study, the distance from the surface of
the RAID to the plane of the steering wheel rim was set to
13 mm to achieve ‘worst case’ response. Positions closer
to the steering wheel were not investigated. The steering
wheel was oriented as shown in Figure 4 with the RAID
perpendicular to the air bag tear seam. As with the SAE
arm tests, a five-axis load cell was located behind the
steering wheel to measure reaction forces. The time of air
bag cover opening was determined using break wires over
the tear seam. In addition, a backstop with foam padding
was used to stop the RAID after the test.
Figure 7. Side View o/RAID.
1634
q Peak Tank Pressure
b
15 EXPERIMENTAL RESULTS
Four OEM air bagst hat are representativeo f a wide
range of air bag aggressivities in the current automobile
fleet were used in the testing. These air bag systems, in
order of decreasing aggressivity identified in previous
RAID testing [Kuppa 19971, are denoted System K,
System H, System J, and System L. The air bags were
mounted in original equipment steering wheels
appropriate for the air bag tested. Inflator performance of
each air bag system from tank testing (60 L tank) is
shown in Figure 8. Tank tests for System J are not
available. Tests on the remaining inflators confirm the
ordering of aggressivity suggested in the RAID testing.
The SystemK inflator is very aggressivew ith a high peak
pressure and a high pressure onset rate. During this
study, several System K air bags burst around the vent
holes during deployment. System H inflators are also
very aggressive with peak pressures slightly lower than
those seen in System K inflators but with a high pressure
onset rate. The System L inflator is relatively nonaggressivew
ith a very low peak pressurea nd onsetr ate.
12
P
E
7s
9&
8
0
6;
t
ap !
3
System K System H System L
Figure 8. Static Tank Pressure and Pressure Slope
Curves - Values Based on a 60 L Tank, Pressure Slopes
Derivedporn Maximum IO ms Values.
There are significant differences in the air bag
modules, especially the location of the module tear seam
as shown in Figure 9. The tear seams for the System K
and System L modules are approximately mid-way
between the top and the bottom of the module. In
contrast, the System H module has a very large and heavy
flap with a low tear seam. This large flap has been found
to provide some protection during the initial air bag
deployment to cadaveric arms under air bag deployment
[Bass 19971. System J has a relatively small vertically
oriented tear seam with wide side flaps. The air bags are
all similar in height and width, and the steering wheels are
similar in dimension. Only the System J air bag is
untethered.
1635
System K System L
System J
Figure 9. Sketches of Air Bag Module Covers Indicating the Tear Patterns.
Table 3. Characteristics of Air Bag Modules (All Measurements in mm)
Air Bag System System H System L System K System J I fl
Horizontal width of module
at seam
203
Distance from top of
module to seam
Thickness of flaps
Vertical height of air bag
13x
3.2
686
Horizontal width of air bag 686 635 I 660 I 63.5
Number of tethers I 4 2 I 3 I none
Length of tethers I 267 279 I 318 I -_
;
23X 200
91 78
1 I
108
1636
System K, System H, System J, and System L air
bags were each tested twice with the SAE arm mounted
on the test fixture used in previous cadaveric tests. In
addition to the fixture tests, one System H air bag and one
System L air bag were deployed into the SAE arm
mounted on a 5* Percentile Hybrid III dummy. For the
RAID, one test was performed using each of the air bag
systems in this study. In addition to these tests, several
repeatability tests were performed with the System K and
System J air bags.
A typical deployment for both test series begins with
a bulge in the air bag module following air bag initiation.
Then, the air bag deploys through a scored tear seam
oriented perpendicular to the forearm. In the initial stages
of air bag inflation with the SAE arm, there is no
significant humerus motion, and the forearm begins to
rotate about the elbow until it reaches the joint stop.
After the elbow reaches the joint stop, the humerus begins
rotating toward the center of the steering wheel. This
continues until the SAE arm hits the dummy in the
Hybrid III tests or the backstop in the fixture tests. For the
RAID, the deployment rotates the arm until the arm
contacts the padded backstop. For the SAE arm mounted
to the Hybrid III, there is no substantial shoulder
movement until the air bag deploys into the dummy chest.
Moment time histories from both test devices suggest that
the greatest forces on the forearm occur during the air bag
punch-out and shortly thereafter.
All air bags deployed normally except for one of the
System K air bags in the SAE arm testing. As seen in a
previous cadaveric test series [Bass 19971, the System K
air bag suffered large tears during the deployment
originating at the reinforced seam around the peripheral
vent holes. In spite of these holes, the air bag appeared
to inflate fully.
Peak resultant forearm bending moments for both
the SAE arm and the RAID are presented in Error!
Reference source not found.. For repeated tests within
each air bag system using the SAE arm, these moments
are consistent, showing a maximum difference of
approximately 10%. Peak humerus axial loads for the
SAE arm are not as consistent since they are associated
with the details of the air bag/elbow joint stop interaction
at times greater than injury times identified in cadaveric
tests. So, these peak humerus axial loads are not generally
relevant to primary contact injuries.
Peak moment values for the RAID are much larger
than those measured using the SAE arm. This is likely
the result of the RAID having greater mass and moments
of inertia than the SAE arm. In addition, the ordering of
aggressivity quantified using peak moments of the
System K and System H air bags is reversed in the RAID
1637
from that found using the SAE arm. This is likely the
result of a heavier air bag module cover with a lower air
bag seam than the rest of the test devices. As the SAE
arm testing placed the distal third of the forearm on the
air bag tear seam while the RAID maintained uniform
radial placement with respect to the steering wheel, the
larger, lower flap of the System H air bag tends to
increase peak moments for the RAID relative to the SAE
arm. However, the peak moments derived from testing
using the SAE arm and testing using the RAID, compared
in Figure 10, show a correlation coefficient of 0.90
indicating similar peak moment response.
140
p 120.
t
E loo-
!
a
P
80.
z
$ 6c- A
Y
B 40- A
E
4:
4 20.
fn
A
A
04
200 300 400 500 600
RAID Peak Bending Moment (N-m)
i
700
Figure IO. Peak Forearm Moments of the SAE Arm and
ZUZD.
Forearm moment time histories for each of the
systems tested are plotted in Figure II for the SAE arm
and in Figure 12 for the RAID. As expected, the peak
SAE arm forearm bending response of System K and
System H is significantly greater than that seen in System
J or System L. Large peak moments after 15 ms are
associated with the SAE arm elbow reaching the joint
stops. Interestingly, the peak forearm moments from the
System K tests are much earlier than those seen in the
System H tests. High-speed video analysis indicates that
while peak bending moments occur during module
cover/arm interactions for System K, the peak moments
for System H occur after the time that the arm interacts
with the module cover. This indicates that while the
module cover may play a role in injuries, module cover
interaction may not be necessary for such primary contact
injuries.
For the RAID, the timing of forearm moment peaks
is similar to those found with the SAE arm. The System
K air bag has the earliest peak, and the System H air bag
has the latest peak moments. Though the order of peak
forearm moment was switched between System L and
System J air bags for the RAID and SAE arm, the timing
of these peak moments was similar. Both the SAE arm
and the RAID show peak moments for System H after the
time of arm/module cover interaction.
In contrast, with the System K air bag, the second
moment peak appears later for the SAE arm than for the
RAID. Because the acceleration of the SAE arm is much
greater than that of the RAID, the RAID is closer to the
inflator when the air bag emerges from the module cover,
producing earlier peak moments. The kinematics of the
SAE arm appear to be more consistent with cadaveric test
results. In addition, the SAE arm moment peaks generally
maintain the order of aggressivity found in previous
cadaveric testing.
As expected from the inertial properties, the peak
accelerations shown in Error! Reference source not
found. using the SAE arm are substantially larger than
the peak accelerations found using the RAID even though
the accelerometers were placed in similar locations. The
RAID has a 40% greater forearm/hand mass and nearly
ten times the forearm lateral moment of inertia. For the
SAE arm, the accelerations generally maintain the order
of aggressivity found in previous cadaver tests. The
relatively aggressive System K air bag demonstrated over
twice the peak forearm acceleration than the other three
air bag systems tested. The System H and System J air
bags showed comparable peak accelerations; however,
the more aggressive System H air bag delivered
approximately 10% more peak impulse to the distal
forearm than the System J air bag during the first 15 ms
of deployment. The similarity of peak accelerations with
dramatically different peak moments may be accounted
for by differences in air bag deployments between System
H and System J. From high-speed video, the System J air
bag appears to deploy in a smaller forearm area than do
the System H air bags. One likely source of this
difference is the lack of tethers in the System J air bag.
This concentration of air bag deployment may lead to
increased risk of fracture relative to a tethered bag. In
addition, the System H air bag deploys generally more
distally than the System J air bag when accounting for the
difference in SAE forearm position with respect to the
steering wheel. This effect is not present in the RAID
tests as the test device was not adjusted radially to
account for the differences in air bag tear seam location.
The less aggressive System L air bag demonstrated peak
accelerations and impulses that were substantially lower
than the other air bag systems.
Table 4. SAE Arm and RAID Peak Response Data
istal
cccl.
SAEarm 450 187 208 137
.(g’s) RAID 137 183 65 57
Peak
Humerus SAE Arm 2 110 1660 1680 940
lkxial I
ID NA NA NA NA
Measured shear loads in the SAE dummy forearm
were relatively low, under 800 N for all tests. Such shear
loads are unavailable in RAID instrumentation. These
low forearm shear loads are likely the result of the center
of pressure of the air bag deployment being close to the
center of the load cell.
1638
$ 80
E 0
= 60
.-E
z
; 40
10 15 20 25
Elapsed Time from Airbag Deployment (ms)
Figure 11. SAE Arm Midrhaft Forearm Resultant Bending Moment (Ail Signals Filtered to SAE CFC-600).
800
700
600
2P 500
s
6 400
3
P
z 300
d
200
100
0 P
-System K
-System H
- - . System J
--- System L
10 15 20
Elapsed Time from Airbag Deployment (ms)
Figure 12. RAID Midshaft Forearm Resultant Bending Moment (AI1 Signals Filtered to SAE CFC-600).
1639
SAE elbow flexion is shown in Figure 13 under bending moments entering the joint stop. On the other
System H air bag deployment for typical dummy and hand, all dummy tests see the elbow reach the joint stop
cadaver tests. The two tests see peak flexion angles of later than 20 ms from the time of air bag deployment. As
approximately 50’ with similar timing. The minimal this time is much later than the time of primary contact
effect of the soft joint stop is seen in the System H tests. injury as determined in the cadaveric tests, the behavior in
The SAE arm enters the joint stop region of 40’ flexion at the joint stop is not relevant for research into primary
approximately 18 ms and reaches the limits of travel at contact injuries. So, although flexion to simulate a
approximately 23 ms. In contrast, the effect of the joint human elbow is not expected to be biofidelic with the
stop on the SAE arm is seen clearly in the System K air RAID, these results suggest that the lack of a biofidelic
bag deployment. The slope of the flexion is substantially humerus may not detract from use of the RAID as a
larger than that seen in the System H dummy tests, so the diagnostic device for investigation of primary contact air
arm attains larger velocities and hence larger forearm bag injuries.
0 5 10 15 20 25 30
Elapsed lime from Airbag Deployment (ms)
35 40
Figure 13. SAE Arm - System H - Dummy vs. Cadaver Elbow Flexion and Dummy Forearm Moment (Moment Filtered to
SAE CFC-600, Flexion Angles Filtered to SAE CFC-1000).
0 5 10 15 20 25 30 35 40
Elapsed Time from Airbag Deployment (ms)
Figure 14. SAE Arm - System K - Dummy Elbow Flexion and Dummy Forearm Moment (Moment Filtered to SAE CFC-600,
Flexion Angles Filtered to SAE CFC-I 000).
1640
Figure 1.5 shows the similarity of the responses of
System L air bag deployments into the SAE arm with the
arm on the dummy compared to the arm mounted to the
universal joint fixture. This suggests that such fixture
tests are appropriate for simulation of primary contact
arm/air bag interactions. The three tests show resultant
forearm peak moments that are within 14%, and the
timings of the initial peaks are within 2 ms. Similar
repeatability in the resultant forearm moments is seen in
the System H tests. These results provide additional
evidence that the use of the fixed test fixture with the
SAE arm is appropriate for investigation of primary
contact arm/air bag interactions.
In addition, with the System L air bag, we can
separate the effects of arm/flap and arm/air bag
interaction. The first peaks in bending moment are the
result of flap deployment into the arm, ending at
approximately 7 ms as identified from high-speed video
analysis. The second peaks, however, are solely the result
of arm/air bag interaction. These second peaks rival the
first in magnitude for each of the tests and have
substantially greater impulse.
For the RAID, the results of two repeatedt estsu sing
System K air bags are shown in Figure 16. The initial
peak in resultant moment in repeated tests using the
System K air bag shows only 3% difference in value and
0.1 ms difference in peak timing. In addition, the second
peak shows less than 10% difference in value with a 0.5
ms difference in peak timing. Additional repeated tests
with the System J air bag showed good repeatability in
both peak values and timing. So, both the RAID and the
SAE arm showed good overall repeatibility.
60
50
i
-System L -Arm On Dummy
n Fixture (Test 1)
0 5 10 15 20 25
Elapsed Time from Impact (ms)
Figure IS. SAE Arm - Forearm Resultant Moment - Arm
on Dummy vs. Arm on Fixture (Signals Filtered to SAE
CFC-600).
600 __. System K-Test 1 !
0 5 IO 15 20 25 30
Elapsed Time from Airbag Deployment [ms)
Figure 16. RAID Repeatabiiity - Forearm Moment
Response for System JAir Bag Deployment.
INJURY RISK FUNCTIONS
Since the currently reported arm/air bag tests were
performed in nominally the same condition as previous
cadaveric tests [Bass 19971, we can correlate the injury
results from the cadaveric testing with the average peak
forearm bending moments resulting from air bag
deployment into the SAE arm and the RAID. This is
further justified by the strong correlation between the
peak forearm moment response of the SAE arm and the
RAID. For the forearm bending moments, we use all the
tests with the System K, System H, System J, and System
L air bags as seen in Error! Reference source not found..
For the cadaveric forearms, we limit the injury sample to
the 11 small female cadaveric subjects tested with the
same air bags on the SAE arm test fixture. For a model of
fracture/no fracture, the cadaveric response shows
complete separation at an average SAE arm peak forearm
moment value of 61 N-m. If, however, we assume a
polytomous process where the level of fracture in the
cadaveric tests is associated with the average bending
moment for the repeated tests with a given air bag, we
obtain the logistic regression for the probability of either
an ulna or an ulna/radius fracture for the SAE arm as
shown in Figure 27. The result is statistically significant
to p=O.O2T. he regressions uggestsa 50% risk of at least
one fracture at 67 N-m (+/- 13 N-m Standard Deviation)
forearm moment in the SAE arm under the same test
conditions. The risk of both radius and ulna fracture
using the same model for the SAE arm is shown in Figure
18. This curve suggests that there is a 50% risk of both
radius and ulna fracture at 91 N-m (+/- 14 N-m Standard
Deviation) peak forearm bending moment in the SAE
arm. For both logistic risk curves, the one standard
deviation confidence intervals are plotted.
1641
Injury risk functions for the RAID using peak
moment values correlated with cadaver injury data are
shown in Figure 19 and Figure 20. These risk functions
indicate that for 50% risk of ulna or ulna/radius fractures,
the RAID peak forearm moment is 373 N-m (+/- 83 N-m
standard deviation). For 50% risk of fracture of both the
ulna and the radius, the RAID peak forearm moment is
473 N-m (+/- 60 N-m standard deviation). The results are
statistically significant to p = 0.06, and the one standard
deviation confidence intervals are plotted.
Table 5. Test Device Peak Bending Moments vs.
Cadaveric Injuries [Bass 19971.
Average Peak I
These risk functions for forearm fracture can be
analyzed using the available quasistatic ultimate bending
moments for isolated arm bones reported above.
Grouping all the available tests, we obtain a weighted
average value of 39 N-m for ulna ultimate strength.
Carter and Hayes [Carter 19761 suggest dynamic
dependence on strain rate of the form F oc &“06 where
F is a compressive ultimate load and E is the dynamic
strain rate. For our typical dynamic strain rates of 5 per
second, the Carter and Hayes strain rate dependence
results in 53% increase in ultimate strength for dynamic
bending as compared with UVa quasistatic ultimate
strength for the ulna. This is consistent with the
suggestion of Melvin and Evans [Melvin 19851 who
suggest an increase of 50% for dynamic ultimate strength
over quasistatic ultimate strength. Further, Schreiber et al
[Schreiber 19971 report a 68% increase in the dynamic
bending strength of the tibia over quasistatic tests at strain
rates of 5 per second.
So, if we assume 50% increase in the ultimate
strength of the isolated ulna, the dynamic bending
strength of the isolated ulna is approximately 59 N-m. If
we assume that the radius provides some support under
dynamic bending in the region of the distal third of the
forearm, the 50% risk of fracture at SAE dummy forearm
moments of 67 N-m seems quite consistent with the
quasistatic data. In addition, for a pronated subject arm,
we expect a forearm ultimate strength to be less than the
sum of the ultimate strengths of the radius and the ulna.
Dynamic drop tests presented above suggest that there
may be a 30% decrease in dynamic ultimate strength from
impact into a pronated arm as compared with a supinated
arm. If we assume that the ultimate strength of a
supinated forearm is approximately the sum of the
ultimate strength of the radius and ulna, we obtain a
weighted average ultimate forearm bending moment of 73
N-m under quasistatic conditions. Further, if we
compensate this value as above for the increase in
dynamic ultimate strength and for the decrease in ultimate
strength owing to arm pronation, we obtain an ultimate
dynamic bending strength of approximately 76 N-m for
the forearm in a pronated position. This compares well
with the 50% risk SAE arm moment value of 91 N-m for
forearm ulna and radius fractures. Given the nature of the
approximations above, there is a rough correspondence
between quasistatic bending results and the derived risk
functions for the SAE forearm.
Using this simple order of magnitude analysis, it is
clear that the moments measured in the RAID are far
larger than expected in small female human forearms
under primary contact from a deploying air bag.
However, the RAID was designed as a research tool to
investigate air bag aggressivity and primary contact air
bag injuries. As shown above, measurementst aken using
the RAID under air bag deployment can be successfully
correlated with both cadaver injury and more biofidelic
test devices.
0 20 40 60 80 100 120 140
SAE 8th X Female Resultant Forearm Moment (N-m)
Figure 17. SAE Arm - Risk of Ulna or Radius/Ulna
Fracture.
1642
0 20 40 60 80 100 120 140
6AE 5th % Female Resultant Fonann Moment (N-m)
Figure I8. SAE Arm - Risk of Radius and Ulna Fracture.
0 100 200 300 400 500 600 700 800
RAID MaxImum Resultant Forearm Moment (N-m)
Figure 19. RAID - Risk of Ulna or Radius/Ulna Fracture.
0 100 204 300 400 600 600 700 600
RAID Maximum Resultant Foreann Moment (N-m)
Figure 20. RAID - Risk of Radius and Ulna Fracture.
CONCLUSIONS
This study investigatedt he primary contactp haseo f
air bag deployment into dummy upper extremities using
four OEM air bags representativeo f a range of air bag
aggressivities in the current automobile fleet. This
aggressivity may be quantified using forearm moment
responseo f a dummy surrogatei n an appropriatew orstcase
position. Using this measure for primary contact
injuries, this study found the System K air bag and the
SystemH air bag to be relatively more aggressive,t he
System J air bag to be moderately aggressive, and the
System L air bag to be less aggressive.
Maximum momentsa nd accelerationsfo r both test
devices under air bag primary contact occur early during
air bag deployment. However, peak forearm moments
obtained using a System H air bag with the SAE arm
occurred after the time of significant module cover/arm
interaction. So, module cover interaction may not be
necessarfyo r injury with currentO EM air bags.
Both the RAID and the SAE arm were found to be
appropriatefo r examinationo f air bag aggressivityu nder
primary air bag contact. Results from previous cadaveric
tests suggest that primary contact injuries occur very
early, before significant elbow flexion occurs. This is
confirmed with moment and acceleration results from
both the SAE arm andt he RAID. This suggeststh at both
devices can be successfully correlated with cadaver
primary contact injury data.
There is, however, one significant potential caveat
with the use of the RAID for primary contact injuries into
smah female occupants. As the result of a large mass and
lateral moment of inertia, the kinematic responseo f the
RAID is dramatically different from both a human
forearm and the more biofidelic SAE arm. This is seen
1643
clearly in the distal acceleration response of the RAID.
For all air bag systems, the acceleration was substantially
smaller than that seen with either the human forearm or
the SAE arm. So, tbe RAID will not generally be suitable
for the investigation of forearm moment response of
primary contact phenomena that depend on details in
timing of the arm/module cover/air bag interaction. In
addition, for the investigation of later phases of air bag
deployment, flinging, or occupant contact, the SAE arm is
more appropriate since its allowed motions are
approximately anthropomorphic.
A comparison of tests using the SAE arm mounted
to a Hybrid III dummy and the SAE arm mounted to a
universal joint test fixture show that the use of a
translationally fixed furture has minimal effect on forearm
response. So, either the SAE arm or the RAID may be
used in a fixed test furture for experimental convenience
without significant effect on primary contact response.
The dummy forearm moment obtained under air bag
deployment into tbe SAE arm and RAID correlates well
with injury levels observed in cadaveric testing with the
same upper extremity orientation. A logistic injury risk
function was developed for small females in the ‘worstcase’
position using the cadaveric injuries and the dummy
forearm moments. This risk fimction predicts a 50% risk
of ulna fracture at a SAE forearm moment of 67 N-m (+/-
13 N-m standard deviation) or a RAID moment of 373 Nm
(+/- 83 N-m standard deviation). The SAE arm value is
consistent with an extrapolation of quasistatic ultimate
bending strength of the ulna to dynamic conditions. As
the result of differences in mass and moments of inertia,
the moment value in the RAID is not expected to be
similar to those found using a more biofidelic small
female arm. In addition, we find a 50% risk of radius and
ulna fracture at a SAE forearm moment value of 91 N-m
(+/- 14 N-m standard deviation) that is consistent with the
combined bending strength of the radius and ulna in a
pronated position. A similar risk of two forearm fractures
is seen with a RAID peak forearm moment of 473 N-m
(+/- 60 N-m standard deviation).
ACKNOWLEDGEMENTS
The authors gratefully acknowledge the support and
guidance of Nopporn Khaewpong (NHTSA) and Rolf
Eppinger (NHTSA). This work was supported in part by
DOT NHTSA Cooperative Agreement DTNH-22-96Y-
07029, in part by DOT NHTSA Contract Number
DTHN22-92-D-07092, and by the University of Virginia
School of Engineering and Applied Science. In addition,
the authors sincerely thank Honda R&D North America
for the use of the SAE 5” Percentile Instrumented arm.
REFERENCES
[Bass 19971 C.R. Bass, S.M. Duma, J.R. Crandall, R.
Morris, P. Martin, W.D. Pilkey, S. Hurwitz, N.
Khaewpong, R. Eppinger, and E. Sun, The
Interaction of Air Bags With Upper Extremities,
SAE Paper 973324, 41st Stapp Car Crash
Conference, Orlando, Florida, 1997.
[Bass 19981 C.R. Bass, J.R. Crandall, and W.D. Pilkey,
A Fixture for the Investigation of Air Bag
Deployments into Out-of-Position Occupants,
Unpublished Manuscript, 1998.
[Carter 19761 D. Carter and W. Hayes, Bone
Compressive Strength: The Influence of Density
and Strain Rate, Science, 194: 1174, 1976.
[Freedman 19951 E.L. Freedman, M.R. Safian, and
R.A. Meals, Automotive Air Bag-Related Upper
Extremity Injuries: A Report of Three Cases,
Journal of Trauma, 38~577, 1995.
[Huelke 19951 D.F. Huelke, J.L. Moore, T.W.
Compton, J. Samuels, and R. Levine, Upper
Extremity Injuries Related to Air Bag
Deployments, Journal of Trauma, 38:482, 1995.
[Johnston 19971 K.L. Johnston, K.D. Klinich, D.A.
Rhule, and R.A. Saul, Assessing Arm Injury
Potential from Deploying Air Bags, SAE Paper
Number 970400, SAE International Congress
and Exposition, Detroit, MI, 1997.
[Jurist 19771 J. Jurist and A. Folitz, Human Ulnar
Bending Stiffness, Mineral Content, Geometry
and Strength, Journal of Biomechanics, IO:45 5,
1977.
[Kirchoff 19951 R. Kirchhoff, and S.W. Rasmussen,
Forearm Fracture Due to the Release of an
Automobile Air bag, Acta Orthopaedica
Scandinavica, 66:483, 1995.
[Kuppa 19971 SM. Kuppa, C.W. Yeiser, M.B. Oslon, L.
Taylor, R. Morgan, and R. Eppinger, RAID - An
Investigation Tool to Study Air Bag/Upper
Extremity Interactions. SAE Paper Number
970399, SAE International Congress and
Exposition, Detroit, MI, 1997.
[Marco 19961 F. Marco, A. Garcia-Lopez, C. Leon, and
L. Lopez-Duran, Bilateral Smith Fracture of the
Radius Caused by Air Bag Deployment, Journal
of Trauma, 40:663, 1996.
[McConville 19781 J. McConville and L. Laubach,
Anthropometry, in Anthropomeby Source Book,
VoIume I, NASA RP-1024, eds. W. Thornton
and J. Jackson, 1978.
[Melvin 19851 J. Melvin and F. Evans, Extremities;:
Experimental Aspects, in The Biomechanics of
Trauma, eds. A. Nahum and J. Melvin,
Appleton-Century-Crofts, Norwalk, CT, 1985.
1644
[Melvin 19931 J. Melvin, J. Horsch, J. McCleary, L.
Wideman, J. Jenson, and M. Wolanin,
Assessment of Air Bag Deployment Loads, SAE
Paper 933 119, 37ti Stapp Car Crash Conference,
San Antonio, Texas, 1993.
[Roth 19931 T. Roth, and P. Meredith, Hand Injuries
From Inflation of an Air Bag Security System,
Journal OfHandSurgery, 18B:520,1993.
[Saul 19961 R.A. Saul, S.H. Backaitis, M.S. Beebe,
and L. Ore, Hybrid III Dummy Instrumentation
and Assessment of Arm Injuries During Air Bag
Deployment, SAE Paper 962417,4Oth Stapp Car
Crash Conference, Albuquerque, New Mexico,
1996.
[Schreiber 19971 P. Schreiber, J. Crandall, T. Micek, S.
Hurwitz, and G. Nusholtz, Static and Dynamic
Bending Strength of the Leg, 1997 International
Ircobi Conference on the Biomechanics of
Impact, Hanover, Germany, September 1997.