II. INJURY RISK CURVES AND PROTECTION REFERENCE VALUES

One of the most difficult problems in determining the benefits of a safety countermeasure is how to translate measurements from a test dummy to humans. For this analysis the agency is using some of the injury curves that it has used in the past and has developed many new injury curves or injury assessment reference values. These will be discussed by body region.

Head

In the "Final Economic Assessment, FMVSS No. 201, Upper Interior Head Protection", NHTSA, June 1995, the agency utilized two different sets of HIC versus injury curves. These were labelled by NHTSA as the Expanded Prasad/Mertz curves and the Lognormal Curves. Only the AIS 4+ curves were developed by Prasad/Mertz, the other curves were developed by NHTSA based on the AIS 4+ Prasad/Mertz curve.

The formulae for the Expanded Prasad/Mertz curves are:

The lognormal curves do not have formulae. Tables II-1 and II-2 give estimated chance of injury for the two sets of curves.

Figure II-1 shows the chance of fatality predicted by the two curves.

These estimated injury levels from a given HIC are used to estimate the injuries caused by air bags for out-of-position occupants and to determine head injury levels when comparing baseline to depowered air bags.

As part of the neck discussion, there are protection reference values for children and adults provided. The protection reference values for the head are close to 1,000 HIC, the FMVSS 208 and FMVSS 213 requirements. Slight differences are recommended for 5th and 95th percentile dummies. HIC of 1,000 is used for 3 and 6 year old dummies, but a HIC of 500 will be used for infants (in this case with the 12-month old dummy). This reflects the skull of infants not having hardened.

Table II-1

Table II-2

Lognormal Curves

Figure II-1

Neck

Based on existing biomechanical literature, the agency developed a new neck injury criterion to deal with this problem of neck injuries for out-of-position occupants. The reader is referred to the docketed paper "Techniques for Developing Child Dummy Protection Reference Values", August 1996, to read about the neck injury criteria. A normalized resultant load plotted against time duration has been developed that takes into account the measurements of neck tension/compression, flexion/extension, shear, and the duration of the pulse in determining the chance of neck injury for a human given these various dummy measurements (see Figure II-2 for an example). Different values relate to different injury levels for children and for adults. A table of neck injury curves was developed for AIS 2 injuries and for AIS 5 injuries. AIS 5 injuries are considered to be equivalent to fatalities for this analysis. The chance of injury and fatality given an overall neck injury risk value is shown in Table II-5.

The protection reference values are shown in Table II-3 and II-4 for children and adults. The values in Table II-3 were mainly developed by NHTSA. The values in Table II-4 are General Motors values and don't necessarily reflect current NHTSA standards or values. For example, the NHTSA standard is 1,000 HIC at 36 ms, while the GM reference values are at 15 ms intervals. The agency recognizes that there are some inconsistencies between the GM 5th percentile female reference values and those NHTSA supports for 6 year old child dummies. The agency is working to resolve these differences.

 

Table II-3

Hybrid III/CRABI Out-of-position Child Protection Reference Values
Dummy Size 12 MO 3 Year 6 Year
HIC(36) 500 1000 1000
Peak Head Acceleration (g)
Head Angular Velocity I (r/s)
Head Angular Acceleration I (r/s2)		 80
37
<2500		 80
34
<2200		 80
33
<2100
Neck Tension/Extension, NTE
Neck Tension/Extension, NTF
Neck Compression/Extension, NCE
Neck Compression/Flexion, NCF		 Ni j 1
and,
Ni j 1-0.02222 t for 0<t 30msec
Chest Acceleration (g) (Spinal)
V*C (m/sec)		 60
1.0		 60
1.0		 60
1.0

 

Neck measurement critical values for child dummies
Dummy Tension (N) Compression (N) Flexion Moment
(N-m)		 Extension Moment
(N-m)
12 month 2000 2000 50 20
3 year 2500 2500 60 30
6 year 3000 3000 70 35

 

Figure II-2

Table II-4

Hybrid III Adult Dummy Protection Reference Values

Component Body Segment Criteria Small Female Mid-size Male Large Male
Head Head HIC 1113, 15ms 1000, 15ms 957, 15ms
Head/Neck
Interface		 Neck Flexion Bending Moment (Nm) 104 190 258


Extension Bending Moment (NM) 31 57 78


Axial Tensile Loading vs. Time Duration (N) 2201 max. Fig.4A2 3300 max. 4052 max.


Axial Compressive loading vs. Time Duration (N) 2668 max.
Fig. 4A3		 4000 max. 4912 max.


Fore/Aft Shear Force vs. Time Duration (N) 2068 max. Fig. 4A4 3100 max. Fig. 4A4 3807 max. Fig. 4A4
Chest Thoracic Organs Resultant Chest Spine Acceleration (Gs) 73, 3ms 60, 3ms 54, 3ms

Thoracic Organs Compressive Deflection Due to Shoulder Belt 41mm 50mm 55 mm

Thoracic Organs Compresssive Deflection Due to Air Bag & Steering Wheel Hub 55 mm 65 mm 72 mm

Viscous Criterion 1 1 1 1
Femur Patella,
Femur,
Pelvis		 Axial Compressive
Femur Load vs. Time
Duration (N)		 6186 max., 9070
max., 		11537 max.,
Knee PCL Tibia to Femur
Translation		 12 mm 15 mm 17 mm
Knee Clevis Tibial
Plateau		 Med/lat. Clevis Comp.
Loading (N)		 2552 4000 4920
Femur, Tibia
Comp. Loading (N) 5104 8000 9840 N


Tibia Index, TI =M/Mc
= P/Pc
where Mc = Critical
Bending Moment, and
Pc = Critical Comp.
Force		 1
115
22.9		 1
225
35.9		 1
307
44.2
Ankle Ankle compressive Loading (N)
4000 inferred

Table II-5

Out of Position Child
Neck Injury Risk

Nij % Injury Risk*
AIS=2 AIS=5
1 2 5
1.07 2.07 10
1.16 2.16 20
1.226 2.226 30
1.278 2.278 40
1.33 2.33 50
1.382 2.382 60
1.434 2.434 70
1.5 2.5 80
1.59 2.59 90
1.66 2.66 95

* Normal distribution assumed
Nij is the overall neck injury risk value

Arms

The agency has also developed protection reference values for arm injuries. These injuries occur mostly to drivers. These are discussed in Section IV.

Chest

For this analysis, a paper has been docketed that analyzes all cadaver data available and compares the probability of injury at the AIS 3+, AIS 4+, and AIS 5+ levels versus chest g's. These data were segregated into a shoulder belted group and an air bag group. The shoulder belted group (80 tests) included 3-point manual lap/shoulder belts, 2-point automatic belts, and 3-point manual belts with air bags. The air bag group (35 tests) included air bags alone and air bags with lap belts. These data were adjusted for the age of occupants found in frontal crashes, rather than relying on the cadaver ages (which average 52 years old).

The age-adjusted results are shown in Figure II-3 and Table II-6 for belt restrained occupants and Figure II-4 and Table II-7 for air bag restrained occupants (without belts). The cadaver data with air bags alone is provided in Table II-8 and Figures II-5 to II-7 show confidence intervals around the AIS 3, AIS 4, and AIS 5 air bag age adjusted curves.

Currently, most restraint systems with lap/shoulder belts and air bags are dominated by the belt system. The belt system takes most of the load before the occupant gets into the air bag. These systems are being changed; some belts are being designed with more elasticity, allowing the occupant to stretch forward and have the air bag take more of the load.

Comparing results in Table II-6 and II-7 shows that the same chest g's for belted cadavers result in a much higher probability of injury than for an air bag cadaver (e.g., at 60 g's, 38.5 percent of belted cadavers would have an AIS 4 or greater injury, whereas about 7.8 percent of air bag only cadavers would have an AIS 4 or greater injury). These curves were the biomechanical basis for the belief that the 80 chest g's level proposed for the air bag alone test was similar or safer than the 60 chest g's level for belted occupants. In all three cases (AIS 3+, AIS 4+ and AIS 5+) the probability of injury for an air bag alone at 80 g's (see Table II-7) is lower than the probability of injury for belts at 60g's (see Table II=6). The closest comparison is for AIS 5+ injuries, the probability of injury for belts at 60g's of 4.3 percent relates to the same probability of injury for air bags at 85 g's. This is why the agency proposed either 80 g's for the air bag alone full barrier test and/or a generic 125 msec sled test. While Tables II-6 and II-7 show the probability of chest injury, based on cadaver testing, these are not necessarily directly applicable to Hybrid III dummy chest g's measurements. In Chapter IV an estimate will be made of the change in chest injuries to occupants from depowering.

The agency examined the 1991-95 NASS CDS files. For each AIS level when the chest injury is the highest or equal to the highest AIS and there is a fatality, then this was counted as a fatality due to a chest injury. The agency found that 85 percent of the AIS 5 chest injuries and 100 percent of the AIS 6 chest injuries, averaging 90 percent of the AIS 5+ chest injuries, resulted in a fatality. Thus, the agency could consider the AIS 5+ curves fairly comparable to fatalities.

Figure II-3

Table II-6

Probability of Injury for Belt Tests

Figure II-4

Table II-7

Probability of Injury for Air Bags

Figure II-5

Figure II-6

Figure II-7

In their January 30, 1997 docket comment, AAMA asserts that the injury risk curves for air bag restrained cadavers are based on only one AIS 3, one AIS 4, and one AIS 5 observation. Attachment 3, Page 10 claims that data exist that demonstrate that chest deflection and viscous criterion are better measures of chest injury risk than chest acceleration.

Response: The curves presented in Figures II-5 to II-7 were derived from data resulting from 35 experiments. The resulting data set (shown in Table II-8) is comprised of 15 AIS(0), 2 AIS(1), 7 AIS(2), 3 AIS(3), 6 AIS(4) and 2 AIS(5) cases. This gives adequate quantity and distribution of injuries to allow a logistic regression analysis to be conducted (using PROC/LOGISTIC of SAS).

It is not clear what data are the basis of AAMA's statement that chest deflection or viscous criterion (v*c) are better measurements of chest injury risk than chest acceleration. AAMA provided no data on this point. NHTSA is aware of only 16 experimental tests, tests conducted for NHTSA< where the thorax was restrained by an air bag and the instrumentation used was capable of simultaneously obtaining both chest deformation and chest acceleration. Therefore, NHTSA believes that these 16 tests form the only legitimate data set from which a comparative analysis of the respective veracity of v*c, deflection and acceleration to predict thoracic injury levels can be conducted. Figures II-8, II-9, and II-10 illustrate results of a logistic regression analysis where each of the three independent variables are used to predict the probability of the resulting thoracic injuries being greater than or equal to an AIS 3 injury. Associated with each analysis is a "p-value" which is the calculated probability that the observed relationship could be found in data where there is no association between the independent variable and the outcome. Chest deflection, with a p-value of 0.57 is the poorest predictor. The next best predictor is v*c with a p-value of 0.175. Chest acceleration (g's), with a p-value of 0.026, indicating that it has only a 2.6 percent chance that its level of predictive capability could be derived from random data is by far the best predictor of injury outcome.

Therefore, while there may be other thoracic loading conditions reported in scientific literature where either v*c or chest deflection predict the probability of thoracic injury better than chest acceleration, the evidence presented here is clear that, when considering the situation where the thorax is restrained solely by an air bag, chest acceleration is the most appropriate independent variable to use when predicting injury outcome.

 
Table II-8

Cadaver Sled Tests with Air Bag Restraints
Used in the Analyses

Reference No. Max. Thor. AIS Age Chest (g's) Chest Defl. V*C
MCW118 0 29 44 0.22 0.47
UH9014 0 31 35 0.19 0.41
[6] 0 57 31

MCW114 0 58 60 0.23 0.55
UH9207 0 25 48 0.1 0.14
MCW124 0 76 18 0.19 0.61
UH9212 0 38 46 0.15 0.31
[7] 0 63 97

[7] 0 67 62

[8] 0 66 39

[8] 0 26 67

[8] 0 26 75

[8] 0 37 63

[8] 0 18 45

[8] 0 31 47

[8] 1 55 44

[6] 1 61 42

MCW127 2 81 21 0.11 0.14
MCW112 2 67 44 0.15 0.33
MCW113 2 64 43 0.32 0.76
[6] 2 63 51

[6] 2 61 25

[8] 2 43 54

[6] 2 57 38

MCW126 3 64 27 0.18 0.35
MCW125 3 75 46 0.26 0.61
[8] 3 65 76

UVA93 4 66 67 0.33 2.05
MCW119 4 71 54 0.24 1.74
UVA96 4 58 111 0.05 0.05
[7] 4 68 95

[7] 4 56 67

[7] 4 66 93

UVA94 5 66 88 0.25 0.48
UVA97 5 67 70 0.13 0.19

See References at end of chapter

FIGURE II-8

FIGURE II-9

FIGURE II-10

 

 

REFERENCES

6. Walsh, M. J. and Kelleher, B.J., "Evaluation of Air Cushion and Belt Restraint Systems in Identical Crash Situations using Dummies and Cadavera (780893)," Twenty-Second Stapp Car Crash Conference, October 1978.

7. Cheng, R., Yang, K. H., Levine, R. S., King, A. I., and Morgan, R. M., "Injuries to the Cervical Spine Caused by a Distributed Frontal Load to the Chest (821155)," Twenty-Sixth Stapp Car Crash Conference, October 1982.

8. Kallieris, D., Mattern, R., Schmidt, G., Klaus, G., "Comparison on Three-Point Belt and Air Bag-Knee Bolster Systems. Injury Criteria and Injury Severity at Simulated Frontal Collisions," International Research Council on Biokinetics of Impact, September, 1982, pp 166-183.

MCW = Medical College of Wisconsin

UH = University of Heidelberg

UVA = University of Virginia