K. H. Yang
Computer Simulation of Occupant Neck Response to Airbag Deployment in Frontal Impacts
B. K. Latouf A. I. King Bioengineering Center and Dept. of Mechanical Engineering, Wayne State University, Detroit, Michigan 48202
A mathematical simulation was performed to study the potential of head and neck injury to an unbelted driver restrained by an airbag. The baseline study represented a 50th percentile male dummy driving in a compact car with the steering wheel perpendicular to the floor. The vehicle was moving at 48 km/hour at the time of impact. Model predictions were compared with sled test results. The data agreed reasonably well. A parametric study was performed to study the effect of changing the steering wheel angle and the size of the airbag. It was found that when the standard 20 degrees angle steering wheel was used, neck joint torques were decreased by 22 percent while the resultant head acceleration increased 41 percent from the base line study. When the vertical dimension of the airbag was reduced by 10 percent, neck joint torques were increased by 14 percent, while head acceleration showed a slight decrease of 9 percent.
Introduction U.S. Federal Motor Vehicle Safety Standard (FMVSS) 208 requires all 1990 passenger cars to be equipped with a passive restraint system for both outboard front-seat occupants. Currently, automatic belts and airbag restraints are the only two systems which appear to comply with this safety standard. The airbag is being installed in increasing numbers in new cars as a supplemental restraint system. Cheng et al. (1982) found severe neck injuries in frontal impacts of unbelted cadavers against a predeployed driver side airbag. Three out of six cadavers sustained what would have been a fatal C1-C2 separation or ring fracture. In a subsequent calculation of neck resultant loads, Yang et al. (1985) proposed a neck injury threshold of 10 kN. The airbags used in these tests were not representative of any known system in production. However, the research indicates that occupant neck load needs to be investigated for production airbag systems. A full-scale sled and barrier crash test is essential in the evaluation of any restraint system design. Because of the high cost and associated instrumentation problems relating to airbag sled testing, the use of mathematical models for simulation of crash victim dynamics has become a popular method of safety research and design. In 1976, King and Chou reviewed several "Gross Motion Simulators" (1976). A recent study by Prasad and Chou surveyed the existing programs capable of doing occupant simulations (1989). The programs either employ principles of Newtonian dynamics or are based on finite element methods. Among the models reviewed, CAL3D (Wright-Patterson Air Force Base, Dayton, OH 45433-6573), a public domain program, is the most popular for occupant crash simContributed by the Bioengineering Division for publication in the JOURNAL OF BIOMECHANICAL ENGINEEEING. Manuscript received by the Bioengineering Division May 24, 1990; revised manuscript received December 21, 1991. Associate Technical Editor: R. L. Spilker.
ulation (1988). Wang has reported several airbag simulations using CAL3D (1988, 1987, 1989, 1990). However, neck forces and moments were not reported in his studies. The purposes of this study were (1) to create a mathematical model which can simulate airbag interaction with an occupant using the publicly available multibody dynamic simulation program CAL3D, (2) to perform a parametric study which addresses occupant neck responses with a standard 20 degrees steering wheel angle and a smaller airbag. CAL3D and Airbag Simulation CAL3D was first developed by the Calspan Corporation for the National Highway Traffic Safety Administration (NHTSA), in 1971-1974 (Fleck 1981). It was later modified in 1975 by Wright Patterson Air Force Base (WPAFB) and also became known as the Articulated Total Body model (ATB). The CAL3D program utilizes three-dimensional Newtonianbased equations in describing the "Gross Motion" of the occupant. The occupant is modeled as a multiply-coupled, rigid body, open tree structure represented by segmented ellipsoids. The vehicle is described as a series of connected planes that can represent the seat, floor, instrument panel, windshield, header and other interior components. Through interactions with the various vehicle contact surfaces, external forces are generated on the body. The relative rotation of the segments between links are resisted by nonlinear torsional springs, and viscous and Coulomb friction dampers. Ellipsoids are used to sense contact by the amount and rate of penetration into the plane from user-specified, force-deflection curves. The resulting equations of motion are solved numerically using a vector exponential integrator. In the early seventies, an airbag algorithm was incorporated into CAL3D (WPAFB 1988). The airbag is represented by an
Journal of Biomechanical Engineering
AUGUST 1992, Vol. 114 / 327-
Downloaded From: https://biomechanical.asmedigitalcollection.asme.org on 01/18/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use
Copyright © 1992 by ASME
SLED ACC. (G'S) Flexion
' Segments
* CAL3D s e g m e n t a n d j o i n t
40
definition
60
80
100
120
140
160
TIME (MSEC)
Fig. 1 CAL3D segment and joint representation of a Hybrid III multisegmented neck model
Fig. 2
Sled acceleration time history
Table 1 Comparison of baseline computer simulation Run A with sled test results
ellipsoid, made of inextensible material. Airbag interaction with occupant ellipsoids is handled by a geometric algorithm. Forces are generated when body parts come into contact with the bag. At each time step, a total decrease in volume of the airbag is calculated based on the interaction of the airbag with all of the contacting ellipsoids. Ideal gas equations are used to calculate the pressure increase in the airbag. The corresponding forces and torques are applied to the various segments based on the increased internal bag pressure and on the surface area of occupant contact. The vehicle is allowed to interact with the airbag by means of reaction panels. The first or primary reaction panel contains the point at which the airbag is attached and from which it is deployed. At time zero the bag is assumed to have no volume and is located at the deployment point on the primary reaction panel. A specified time delay into the crash pulse is required as input from the user in order to represent the sensor activation time. The bag is inflated by using the gas dynamic relationships for choked flow of a gas through a nozzle. The dimensions of the ellipsoidal bag determine the final volume of gas within it. After the bag is fully inflated it is allowed to move dynamically. Until the bag reaches full inflation its orientation with respect to the vehicle is held constant. A spring force is applied to the end of the bag, at the deployment point, on the exterior of the primary reaction panel to keep the bag attached to the vehicle. The simulation program CAL3D along with its post processor program VIEW can be obtained from the Human Engineering Division of Wright-Patterson Air Force Base (Dayton, OH 45433-6573).
Peak Data Values Head Resultant Accelerations (g's) Knee Axial Load (N) Neck Resultant Force (N) Neck flexion Moment (N m)
Sled Test Data
CAL3D Run A
Difference (%) 15
77
88
10000
9162
-8
2400
4742*
98
180
188*
4
Measured at NP joint
MPa while the initial gas supply temperature was 675 K. These are typical values measured in an airbag equipped car. The unbelted dummy was seated at the mid-track position. The airbag restraint was attached to a horizontal steering column to simulate a tilted steering wheel. It also matched the steering column configuration used by Cheng et al. (1982). The deployment started 14 ms after onset of the crash simulation. Thus, the difference between the computer model and the sled test was the airbag configuration. In the sled test, the airbag was predeployed while the model simulated an actual inflation process. The model results, using this single segment neck model, did not match the results from the sled tests.
Modified Simulation. The model was modified by adding a more detailed and improved neck model developed by Deng (1989). Figure 2 shows the Hybrid III neck design and a typical way to specify the segments and joints used in the CAL3D model. All other parameters remained the same as the initial simulation run. After the inclusion of Deng's neck model, the neck resultant Methods torques of CAL3D model compared well with those measured Initial Simulation. The Hybrid III dummy was simulated during the Wayne State sled test (Cheng 1982). Table 1 shows by 17 segments and 16 joints, using the WPAFB data set (1988). a comparison of neck resultant torques, knee loads along the In these data sets, the torso and neck joint spring characteristics axial femur direction, head and chest resultant accelerations were described by restoring torques. The neck consisted of between the model prediction and sled test results. Good agreeonly a single segment which articulated with the upper torso ment was found between the parameters compared. and head. The car interior chosen was a 1988 Ford Topaz Figure 3 is a comparison of the resultant head acceleration (Ford Motor Company, Dearborn, Mich.). The Topaz was between the computer simulation and sled test results. The selected because it was available for measurement in our lab- head acceleration curve correlated well with sled data up to oratory. The sled deceleration pulse used was from a frontal 160 ms. The simulation was performed for 160 ms and conimpact Hybrid III sled test (Fig. 1). This deceleration time tained only the forward excursion response. The rebound was history represented a typical 48 km/hr (30 mph) frontal impact. not fully modeled, but was reflected in the second peak of the The airbag used in the simulation was a standard 272 X 564 sled data. This second peak was caused by contact between X 620 mm bag. The total volume of the gas supply reservoir the dummy and the steel frame of the seat used in the sled was 0.045 cubic meter. The initial gas supply pressure was 0.23 test. That frame was not representative of any production car. 3 2 8 / V o l . 114, AUGUST 1992
Transactions of the ASME
Downloaded From: https://biomechanical.asmedigitalcollection.asme.org on 01/18/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use
AIRBAG
DATAPLOT
CAL3D SLED TEST
TIME
(ms)
XIO
Fig. 3 Comparison of the head resultant acceleration for the baseline computer simulation Run A with sled test results
Fig. 4(a)
Lateral view of the occupant response between 0 and 60 ms
T1UE4USEC)
Fig. 4(b) ms
Oblique view of the occupant response between 0 and 60 ms
|40
Lateral view of the occupant response between 80 and 140
The only draw back of this modified simulation was that the neck force was significantly higher than that from the sled test. However, it was closer to the neck forces calculated from cadaver tests. The results of this modified simulation is referred to as Run A in this paper. Journal of Biomechanical Engineering
Fig. 4(c)
TIUEIUSEC]
Fig. 4(d) ms
MO
Oblique view of the occupant response between 80 and 140
Parametric Studies. Once a reasonable correlation was achieved, between the actual test results and those from the simulation, a parametric study was performed to determine occupant neck response to airbag deployment angle and geometry. The conditions in Run B were identical to those of AUGUST 1992, Vol. 114/329
Downloaded From: https://biomechanical.asmedigitalcollection.asme.org on 01/18/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use
Run A except for the angle of the steering column. Instead of a horizontal deployment angle, Run B used a standard 20degree angle. This simulation was performed to investigate the difference in occupant response between a regular steering wheel and a tilted one. For Run C, a 10 percent reduction in the vertical dimension of the airbag was simulated with the steering wheel vertical, the simulation was performed to investigate the effects of a potential design change to a smaller bag in sports cars. Results Figures 4(a), 4(b), 4(c), and 4(d) show a comparison of occupant kinematics among Runs A, B, and C. Both side and oblique perspectives are presented using the VIEW post processor program. The occupant in all three runs moved approximately the same distance until 60 ms. At 80 ms, it is clear that in both Run A and Run C, the head was above the top of the airbag, thus allowing it to flex more severely and to exert higher torque on the neck. Table 2 contains the resultant peaks of critical occupant response data. Neck flexion moment is the highest in Run C, followed by Run A and Run B. There is a 22 percent reduction from the base line study (Run A) when using a standard steering wheel (Run B) and a 14 percent increase when using a smaller airbag (Run C). Figure 5 shows occupant performance curves of the head pivot (HP) and neck pivot (NP) joint resultant torques for the three simulations. In terms of peak values, the trend is the same. Run C has the greatest torque followed by that of Run A and Run B. It can also be seen in Fig. 5 that the torque restrained by the neck was the longest for Run C followed by
Table 2 Peak values computed from the parametric study Sled Test
Peak Data Values
RunB
88
125
80
806
913
1509
660
Head Resultant Accelerations (g's) HIC value
Run C
Run A
77
62
83
74
67
10000
9162
9027
9948
NP joint
188
147
215
11 joint
117
87
154
12 joint
79
49
117
J3 joint
61
21
95
HP joint
61
60
75
NPonNl
4742
4520
4205
Jl on N2
4618
4403
4104
12 on N3
4490
4273
3996
J3 on N4
4353
4136
3878
Chest Resultant Accelerations (g's) Knee axial load (N) Neck flexion moment (N m)
180
2400
Neck Resultant Force (N)
Run A and Run B. For the neck force, Run B was smaller than the base line Run A by 5 percent. Run C had the smallest neck force which was about 11 percent lower than that in Run A. The femur (knee) load, measured in the axial direction of the femur, for Run B was 2 percent lower than that in Run A while in Run C the femur load was 9 percent higher than that in Run A. The Head Injury Criterion (HIC) for Run B was over the 1000 limit set by FMVSS 208 (Chou 1974). It was 65 percent higher than that in Run A. Discussion Of the many injuries which occur in frontal automotive collisions, severe neck injuries are rare for restrained occupants in non-airbag equipped cars. Wayne State has performed frontal impact sled tests using the airbag restraint system in which severe neck injuries were observed in cadavers (Cheng 1982, Yang 1985). The airbags were pre-deployed and were not representative of any known system in production. In this study, although typical airbag values were used, it is also not representative of any known system. The current CAL3D airbag model limits the user to consider occupant-airbag interaction only after the bag is fully inflated. Before the bag is inflated to its final geometric volume, the pressure within it is assumed to be atmospheric. Thus no forces are generated until the bag reaches its final geometric volume. Any out-of-position occupant problems cannot be modeled. The current CAL3D airbag model also neglects any effect that bag slap and transient pressure has on the occupant. These effects can occur during the deployment process in the presence of an out-of-position occupant. The only airbag forces presently generated on the occupant occur after inflation is complete and when internal bag pressurization exists. Although this limitation prevents the simulation of out-of-position occupants, it does not detract from properly modeling other problems that occur after completion of deployment (Wang 1988, 1987). FMVSS 208 specifies performance requirements for the protection of vehicle occupants in automobile crashes. Currently, it addresses occupant response limits on femur load and head and chest accelerations. Occupant neck response is not a part of FMVSS 208 and presently need not be taken into consideration in automotive crashworthiness design. This study showed that a large neck flexion response, beyond injury tolerance levels, can occur in an airbag restraint system, for an unbelted occupant. It may be argued that the occupant neck response may be implicit in the current FMVSS due to the specification of head and chest accelerations. In the FMVSS 208, the injury criterion for the chest is limited to 60 g's, except for intervals, the cumulative duration of which is not more than 3 ms. For the head injury criterion, the resultant head acceleration shall be such that HIC (Eq. (1)) is less than or equal to 1000 (Chou 1974). Jl JOINT RESULTANT TORQUE
NP JOINT RESULTANT TORQUE
RES. TORQUE (N m)
RES. TORQUE (N m )
:oo
-' -
Rva A
" •*
Run A
—
Run B
—•"•
Run R
—-
Run C
—
Run C
•
iv\ l\ '\
130
too
30
A
_J V %
Computer Simulation of Occupant Neck Response to Airbag Deployment in Frontal Impacts
B. K. Latouf A. I. King Bioengineering Center and Dept. of Mechanical Engineering, Wayne State University, Detroit, Michigan 48202
A mathematical simulation was performed to study the potential of head and neck injury to an unbelted driver restrained by an airbag. The baseline study represented a 50th percentile male dummy driving in a compact car with the steering wheel perpendicular to the floor. The vehicle was moving at 48 km/hour at the time of impact. Model predictions were compared with sled test results. The data agreed reasonably well. A parametric study was performed to study the effect of changing the steering wheel angle and the size of the airbag. It was found that when the standard 20 degrees angle steering wheel was used, neck joint torques were decreased by 22 percent while the resultant head acceleration increased 41 percent from the base line study. When the vertical dimension of the airbag was reduced by 10 percent, neck joint torques were increased by 14 percent, while head acceleration showed a slight decrease of 9 percent.
Introduction U.S. Federal Motor Vehicle Safety Standard (FMVSS) 208 requires all 1990 passenger cars to be equipped with a passive restraint system for both outboard front-seat occupants. Currently, automatic belts and airbag restraints are the only two systems which appear to comply with this safety standard. The airbag is being installed in increasing numbers in new cars as a supplemental restraint system. Cheng et al. (1982) found severe neck injuries in frontal impacts of unbelted cadavers against a predeployed driver side airbag. Three out of six cadavers sustained what would have been a fatal C1-C2 separation or ring fracture. In a subsequent calculation of neck resultant loads, Yang et al. (1985) proposed a neck injury threshold of 10 kN. The airbags used in these tests were not representative of any known system in production. However, the research indicates that occupant neck load needs to be investigated for production airbag systems. A full-scale sled and barrier crash test is essential in the evaluation of any restraint system design. Because of the high cost and associated instrumentation problems relating to airbag sled testing, the use of mathematical models for simulation of crash victim dynamics has become a popular method of safety research and design. In 1976, King and Chou reviewed several "Gross Motion Simulators" (1976). A recent study by Prasad and Chou surveyed the existing programs capable of doing occupant simulations (1989). The programs either employ principles of Newtonian dynamics or are based on finite element methods. Among the models reviewed, CAL3D (Wright-Patterson Air Force Base, Dayton, OH 45433-6573), a public domain program, is the most popular for occupant crash simContributed by the Bioengineering Division for publication in the JOURNAL OF BIOMECHANICAL ENGINEEEING. Manuscript received by the Bioengineering Division May 24, 1990; revised manuscript received December 21, 1991. Associate Technical Editor: R. L. Spilker.
ulation (1988). Wang has reported several airbag simulations using CAL3D (1988, 1987, 1989, 1990). However, neck forces and moments were not reported in his studies. The purposes of this study were (1) to create a mathematical model which can simulate airbag interaction with an occupant using the publicly available multibody dynamic simulation program CAL3D, (2) to perform a parametric study which addresses occupant neck responses with a standard 20 degrees steering wheel angle and a smaller airbag. CAL3D and Airbag Simulation CAL3D was first developed by the Calspan Corporation for the National Highway Traffic Safety Administration (NHTSA), in 1971-1974 (Fleck 1981). It was later modified in 1975 by Wright Patterson Air Force Base (WPAFB) and also became known as the Articulated Total Body model (ATB). The CAL3D program utilizes three-dimensional Newtonianbased equations in describing the "Gross Motion" of the occupant. The occupant is modeled as a multiply-coupled, rigid body, open tree structure represented by segmented ellipsoids. The vehicle is described as a series of connected planes that can represent the seat, floor, instrument panel, windshield, header and other interior components. Through interactions with the various vehicle contact surfaces, external forces are generated on the body. The relative rotation of the segments between links are resisted by nonlinear torsional springs, and viscous and Coulomb friction dampers. Ellipsoids are used to sense contact by the amount and rate of penetration into the plane from user-specified, force-deflection curves. The resulting equations of motion are solved numerically using a vector exponential integrator. In the early seventies, an airbag algorithm was incorporated into CAL3D (WPAFB 1988). The airbag is represented by an
Journal of Biomechanical Engineering
AUGUST 1992, Vol. 114 / 327-
Downloaded From: https://biomechanical.asmedigitalcollection.asme.org on 01/18/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use
Copyright © 1992 by ASME
SLED ACC. (G'S) Flexion
' Segments
* CAL3D s e g m e n t a n d j o i n t
40
definition
60
80
100
120
140
160
TIME (MSEC)
Fig. 1 CAL3D segment and joint representation of a Hybrid III multisegmented neck model
Fig. 2
Sled acceleration time history
Table 1 Comparison of baseline computer simulation Run A with sled test results
ellipsoid, made of inextensible material. Airbag interaction with occupant ellipsoids is handled by a geometric algorithm. Forces are generated when body parts come into contact with the bag. At each time step, a total decrease in volume of the airbag is calculated based on the interaction of the airbag with all of the contacting ellipsoids. Ideal gas equations are used to calculate the pressure increase in the airbag. The corresponding forces and torques are applied to the various segments based on the increased internal bag pressure and on the surface area of occupant contact. The vehicle is allowed to interact with the airbag by means of reaction panels. The first or primary reaction panel contains the point at which the airbag is attached and from which it is deployed. At time zero the bag is assumed to have no volume and is located at the deployment point on the primary reaction panel. A specified time delay into the crash pulse is required as input from the user in order to represent the sensor activation time. The bag is inflated by using the gas dynamic relationships for choked flow of a gas through a nozzle. The dimensions of the ellipsoidal bag determine the final volume of gas within it. After the bag is fully inflated it is allowed to move dynamically. Until the bag reaches full inflation its orientation with respect to the vehicle is held constant. A spring force is applied to the end of the bag, at the deployment point, on the exterior of the primary reaction panel to keep the bag attached to the vehicle. The simulation program CAL3D along with its post processor program VIEW can be obtained from the Human Engineering Division of Wright-Patterson Air Force Base (Dayton, OH 45433-6573).
Peak Data Values Head Resultant Accelerations (g's) Knee Axial Load (N) Neck Resultant Force (N) Neck flexion Moment (N m)
Sled Test Data
CAL3D Run A
Difference (%) 15
77
88
10000
9162
-8
2400
4742*
98
180
188*
4
Measured at NP joint
MPa while the initial gas supply temperature was 675 K. These are typical values measured in an airbag equipped car. The unbelted dummy was seated at the mid-track position. The airbag restraint was attached to a horizontal steering column to simulate a tilted steering wheel. It also matched the steering column configuration used by Cheng et al. (1982). The deployment started 14 ms after onset of the crash simulation. Thus, the difference between the computer model and the sled test was the airbag configuration. In the sled test, the airbag was predeployed while the model simulated an actual inflation process. The model results, using this single segment neck model, did not match the results from the sled tests.
Modified Simulation. The model was modified by adding a more detailed and improved neck model developed by Deng (1989). Figure 2 shows the Hybrid III neck design and a typical way to specify the segments and joints used in the CAL3D model. All other parameters remained the same as the initial simulation run. After the inclusion of Deng's neck model, the neck resultant Methods torques of CAL3D model compared well with those measured Initial Simulation. The Hybrid III dummy was simulated during the Wayne State sled test (Cheng 1982). Table 1 shows by 17 segments and 16 joints, using the WPAFB data set (1988). a comparison of neck resultant torques, knee loads along the In these data sets, the torso and neck joint spring characteristics axial femur direction, head and chest resultant accelerations were described by restoring torques. The neck consisted of between the model prediction and sled test results. Good agreeonly a single segment which articulated with the upper torso ment was found between the parameters compared. and head. The car interior chosen was a 1988 Ford Topaz Figure 3 is a comparison of the resultant head acceleration (Ford Motor Company, Dearborn, Mich.). The Topaz was between the computer simulation and sled test results. The selected because it was available for measurement in our lab- head acceleration curve correlated well with sled data up to oratory. The sled deceleration pulse used was from a frontal 160 ms. The simulation was performed for 160 ms and conimpact Hybrid III sled test (Fig. 1). This deceleration time tained only the forward excursion response. The rebound was history represented a typical 48 km/hr (30 mph) frontal impact. not fully modeled, but was reflected in the second peak of the The airbag used in the simulation was a standard 272 X 564 sled data. This second peak was caused by contact between X 620 mm bag. The total volume of the gas supply reservoir the dummy and the steel frame of the seat used in the sled was 0.045 cubic meter. The initial gas supply pressure was 0.23 test. That frame was not representative of any production car. 3 2 8 / V o l . 114, AUGUST 1992
Transactions of the ASME
Downloaded From: https://biomechanical.asmedigitalcollection.asme.org on 01/18/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use
AIRBAG
DATAPLOT
CAL3D SLED TEST
TIME
(ms)
XIO
Fig. 3 Comparison of the head resultant acceleration for the baseline computer simulation Run A with sled test results
Fig. 4(a)
Lateral view of the occupant response between 0 and 60 ms
T1UE4USEC)
Fig. 4(b) ms
Oblique view of the occupant response between 0 and 60 ms
|40
Lateral view of the occupant response between 80 and 140
The only draw back of this modified simulation was that the neck force was significantly higher than that from the sled test. However, it was closer to the neck forces calculated from cadaver tests. The results of this modified simulation is referred to as Run A in this paper. Journal of Biomechanical Engineering
Fig. 4(c)
TIUEIUSEC]
Fig. 4(d) ms
MO
Oblique view of the occupant response between 80 and 140
Parametric Studies. Once a reasonable correlation was achieved, between the actual test results and those from the simulation, a parametric study was performed to determine occupant neck response to airbag deployment angle and geometry. The conditions in Run B were identical to those of AUGUST 1992, Vol. 114/329
Downloaded From: https://biomechanical.asmedigitalcollection.asme.org on 01/18/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use
Run A except for the angle of the steering column. Instead of a horizontal deployment angle, Run B used a standard 20degree angle. This simulation was performed to investigate the difference in occupant response between a regular steering wheel and a tilted one. For Run C, a 10 percent reduction in the vertical dimension of the airbag was simulated with the steering wheel vertical, the simulation was performed to investigate the effects of a potential design change to a smaller bag in sports cars. Results Figures 4(a), 4(b), 4(c), and 4(d) show a comparison of occupant kinematics among Runs A, B, and C. Both side and oblique perspectives are presented using the VIEW post processor program. The occupant in all three runs moved approximately the same distance until 60 ms. At 80 ms, it is clear that in both Run A and Run C, the head was above the top of the airbag, thus allowing it to flex more severely and to exert higher torque on the neck. Table 2 contains the resultant peaks of critical occupant response data. Neck flexion moment is the highest in Run C, followed by Run A and Run B. There is a 22 percent reduction from the base line study (Run A) when using a standard steering wheel (Run B) and a 14 percent increase when using a smaller airbag (Run C). Figure 5 shows occupant performance curves of the head pivot (HP) and neck pivot (NP) joint resultant torques for the three simulations. In terms of peak values, the trend is the same. Run C has the greatest torque followed by that of Run A and Run B. It can also be seen in Fig. 5 that the torque restrained by the neck was the longest for Run C followed by
Table 2 Peak values computed from the parametric study Sled Test
Peak Data Values
RunB
88
125
80
806
913
1509
660
Head Resultant Accelerations (g's) HIC value
Run C
Run A
77
62
83
74
67
10000
9162
9027
9948
NP joint
188
147
215
11 joint
117
87
154
12 joint
79
49
117
J3 joint
61
21
95
HP joint
61
60
75
NPonNl
4742
4520
4205
Jl on N2
4618
4403
4104
12 on N3
4490
4273
3996
J3 on N4
4353
4136
3878
Chest Resultant Accelerations (g's) Knee axial load (N) Neck flexion moment (N m)
180
2400
Neck Resultant Force (N)
Run A and Run B. For the neck force, Run B was smaller than the base line Run A by 5 percent. Run C had the smallest neck force which was about 11 percent lower than that in Run A. The femur (knee) load, measured in the axial direction of the femur, for Run B was 2 percent lower than that in Run A while in Run C the femur load was 9 percent higher than that in Run A. The Head Injury Criterion (HIC) for Run B was over the 1000 limit set by FMVSS 208 (Chou 1974). It was 65 percent higher than that in Run A. Discussion Of the many injuries which occur in frontal automotive collisions, severe neck injuries are rare for restrained occupants in non-airbag equipped cars. Wayne State has performed frontal impact sled tests using the airbag restraint system in which severe neck injuries were observed in cadavers (Cheng 1982, Yang 1985). The airbags were pre-deployed and were not representative of any known system in production. In this study, although typical airbag values were used, it is also not representative of any known system. The current CAL3D airbag model limits the user to consider occupant-airbag interaction only after the bag is fully inflated. Before the bag is inflated to its final geometric volume, the pressure within it is assumed to be atmospheric. Thus no forces are generated until the bag reaches its final geometric volume. Any out-of-position occupant problems cannot be modeled. The current CAL3D airbag model also neglects any effect that bag slap and transient pressure has on the occupant. These effects can occur during the deployment process in the presence of an out-of-position occupant. The only airbag forces presently generated on the occupant occur after inflation is complete and when internal bag pressurization exists. Although this limitation prevents the simulation of out-of-position occupants, it does not detract from properly modeling other problems that occur after completion of deployment (Wang 1988, 1987). FMVSS 208 specifies performance requirements for the protection of vehicle occupants in automobile crashes. Currently, it addresses occupant response limits on femur load and head and chest accelerations. Occupant neck response is not a part of FMVSS 208 and presently need not be taken into consideration in automotive crashworthiness design. This study showed that a large neck flexion response, beyond injury tolerance levels, can occur in an airbag restraint system, for an unbelted occupant. It may be argued that the occupant neck response may be implicit in the current FMVSS due to the specification of head and chest accelerations. In the FMVSS 208, the injury criterion for the chest is limited to 60 g's, except for intervals, the cumulative duration of which is not more than 3 ms. For the head injury criterion, the resultant head acceleration shall be such that HIC (Eq. (1)) is less than or equal to 1000 (Chou 1974). Jl JOINT RESULTANT TORQUE
NP JOINT RESULTANT TORQUE
RES. TORQUE (N m)
RES. TORQUE (N m )
:oo
-' -
Rva A
" •*
Run A
—
Run B
—•"•
Run R
—-
Run C
—
Run C
•
iv\ l\ '\
130
too
30
A
_J V %
