IV. SYNTHESIS OF DATA AND BENEFIT ESTIMATES

In this section, data from the previous sections are brought together in an attempt to determine what are the fatality benefits and disbenefits of depowering air bags. Injury estimates for minor to moderate injuries (AIS 1 and 2 injuries), with the exception of driver arm injuries, have not been entered into the analysis. It is believed that depowering air bags would have significant benefits for minor injuries, but these could not be quantified. Benefits accrue to children and adults who are too close to the air bag currently. Dis-benefits occur at high speeds with unbelted occupants interacting with air bags which have less restraint capability due to depowering. Test results from the 20-35 percent depowered systems are lumped together as an average of vehicles potentially affected by the generic sled test and 80 g's alternative. Based on the AAMA statements, it is assumed that 31 percent of the fleet could be depowered to these level under the 80 g's alternative, while all of the fleet could be depowered using the generic sled test alternative. Based on the AAMA statements, it is assumed that depowering will occur in the 20 to 35 percent range. Under the 80 g's alternative, it is believed that at levels above 35 percent depowering, HIC or other criteria could not be met in an unbelted 30 mph test with the depowered system. Modelling discussed in the PRE showed significant chest injury potential with depowering levels of 40 percent. Depowered levels above 35 percent may be achievable for some models under the generic sled pulse alternative. A higher level of benefits for children might be achievable by depowering some models more, but this would also increase disbenefits for that model. Thus, the analysis only uses test results from depowered air bags in the 20 to 35 percent range. The agency does not believe that manufacturers would depower all air bags to the extent permitted by the amendment. Instead the agency believes that the manufacturers would depower only to the extent needed to address the child and adult fatality problem and preserve unbelted occupant protection to the extent possible. Many caveats and assumptions are needed, because there are many unknowns. This section presents an estimate of what these benefits and dis-benefits might be.

Benefits for children

The benefits for air bags being de-powered are applied to the estimated number of children that could be killed by air bags. If nothing changes (air bag design, belt use, and the use of the right front seat by children are the biggest factors), there will be an estimated 140 fatally injured children by air bags over the lifetime of each model year in which all vehicles are supplied with air bags. Of these, 33 would be infants in rear-facing child restraints. The agency has performed some baseline testing with rear-facing child restraints in front of air bags. In certain cases, with the vehicle seat all the way forward, very high HIC or chest g's are recorded, which could result in fatal injuries to an infant (see Table III-1). In the crashes the agency has investigated of fatal injuries to rear-facing infants, the injury mechanism is almost always skull fractures. The agency does not at this time have a baseline versus depowered air bag set of data for rear-facing child restraints to estimate potential benefits of depowering. The depowered air bags provided to the agency by the manufacturers were either top-mounted bags or other designs that did not result in large HICs for rear-facing child restraint occupants. Thus, these designs were not representative of the fatality problems seen in the real world and testing with depowered systems would not indicate whether depowering could reduce fatalities for infants.

One way to try to extrapolate results from data available is to examine other out-of-position tests. In some out-of-position 3-year-old and 6-year-old dummy tests, the child dummy's head was in direct contact with the air bag cover, or in very close approximation to the air bag cover. These cases were for model B-94 for the 3 year old and 6 year old dummies in Position 2, and for model I-96 for the 6 year old dummy in Position 2 (see Table III-3 and III-5). In these cases, depowering did reduce HIC substantially, but did not consistently bring HIC below the infant injury reference value of 500 HIC, except in the case of 60 percent depowering. Furthermore, it is not known what the 500 HIC for an infant means in terms of injury. While it is possible that depowering could reduce some potentially fatal cases to survivable injury cases, the agency believes depowering will have a minimal impact on the number of infant fatalities in rear-facing child restraints.

An estimated 107 fatalities per year would occur to mostly out-of-position children (not rear-facing) when all light vehicles have air bags. Position 1, 2, and 3 tests have been performed on three different vehicles' air bags with baseline and depowered systems. The depowered systems used for this analysis for the 80 g's and generic sled alternatives are in the 18 to 30 percent depowered range. Model B-94 60 percent depowered tests are not used for estimating benefits in this FRE. Most of the fatalities for forward facing children are neck injuries and head injuries. These data have been run through the new algorithm for neck injury to determine the percent chance of fatality; in addition the percent chance of fatality from HIC has been included (see Table IV-1). The neck injury data is the determinant for the most part of the percent chance of fatality. Most of the children killed have been in the 5 to 7 age group, thus results from the 6 year old child dummy are most representative. Positions 1 and 2 appear to be the most likely in the fatal cases the agency has investigated, the children have started on the seat and none have been known to fall onto the floor before being struck by the air bag. Thus, Position 3 is not used in the calculations. Model D-96 has an estimated very low chance of fatality and is probably not representative of vehicles in which fatalities have occurred (none have occurred in this particular model).

Table IV-1

Percent Chance of Fatality
Forward Facing Children Out of Position Testing

6-year old Position 1 Position 2 Position 3
B-94 Baseline 100% 100% 95%
B-94 D-30% 100 95 5
B-94 D-60% 0.95 1.1 0.5
I-96 Baseline 100 83 94
I-96 D-23% 25 5 1.6
I-96 D-38% 3.5
D-96 Baseline 1.8 - 5.7 4.5 Not tested
D-96 D-18% 1.4 1.2 Not tested

Thus, an average of models B-94 and I-96 in positions 1 and 2 will be used to estimated the potential benefits for depowering air bags on fatalities of children. This implicitly assumes that model B represents 50 percent of the child fatalities and model I represents the other 50 percent and half of the fatalities occur in Position 1 and the other half in Position 2.

The agency has no way of knowing which vehicles would be included in the 31 percent of the fleet that could be depowered under the 80 g's alternative. The impact on safety is very different depending upon whether the depowered models were like model B, like model I, or an equal weighting of the two models. For air bag B-94 (comparing the baseline bag to a 30 percent de-powered bag) the risk of fatality for Position 1 stayed at 100 percent for both bags; for Position 2 the risk decreased from 100 percent to 95 percent (5 percent effectiveness). For air bag I-96 (23 percent de-powered) the risk of fatality for Position 1 decreased from 100 percent to 25 percent (75 percent effectiveness); for Position 2 the risk of fatality decreased from 83 percent to 5 percent (94 percent effectiveness). Averaging the 4 effectiveness percentages together, the risk of fatality dropped by 43.5 percent. Based on the estimated 107 fatalities that would occur if all vehicles had air bags, a 43.5 percent reduction in fatalities (an equal weighting between models B and I) for 31 percent of the fleet would lead to 14 lives saved for children (107 x .435 x .31). If all of the 31 percent of the fleet that was depowered were represented by model I, the highest estimate of lives saved under the 80 g's alternative would be 28 (107 x [.94+.75]/2 x .31).

For the generic sled test alternative, assuming the B-94 model is depowered 30 percent and the I-96 model is depowered 23 percent, the benefits would be 47 child lives saved (.435 x 107 = 47).

Sensitivity analyses could be performed on how to assign the effectiveness estimates in Table IV-1 to the fleet of vehicles and the fatalities occurring today. If 90 percent of all vehicles were represented by B-94, then depowering by the 20-35 percent level would have only a small impact on child fatalities. Similarly, if more vehicles were represented by I-96, more lives would be saved.

Thus, of the estimated 140 children that could be killed by air bags, depowering by an average of 20 to 35 percent for 31 percent of the fleet, the level assumed to be allowed by the 80 g's alternative, is estimated to save 14 lives. Depowering all of the fleet under the generic sled pulse could save 47 lives.

The agency has done testing on the passenger side with current air bags showing that the dummy measurements decrease as the dummy is moved further away from the air bag (see Table III-30). In the one air bag model tested, dummy readings dropped dramatically when the dummy was positioned 7 inches or further from the air bag cover. [Note that the passenger side air bag is generally much more aggressive than the driver side air bag] With a depowered system, the distance from the air bag in which an occupant could be seriously injured would be significantly reduced. This leads the agency to believe that with depowered air bags there should be almost no chance of fatality to a belted child sitting back from the air bag and a large benefit for those moderately out-of-position, while providing less benefit for those within a few inches of the air bag cover upon deployment.

The Impacts of Depowering on Adults

Fatality Benefits

The number of drivers that could be saved by depowering the air bag can not be estimated from the test data currently available. However, in Chapter V an estimate is made of the potential benefits of depowering using some real world data from Australia. In Chapter I, it was estimated that 25 driver fatalities and 7 adult passenger fatalities could occur per year when all vehicles are equipped with air bags. Table III-8 shows that depowering reduced almost all injury measures for out-of-position drivers. However, driver fatalities have occurred with injuries to different body regions ( chest, neck, head). Because fatality curves versus injury measurements for adults are not available for all of these body regions benefits cannot be estimated. Certainly, depowering would affect a large portion but not all of these fatalities. Again, the distance from the air bag at the time of deployment is critical. All things being equal, belted occupants are less likely to move as far forward during pre-crash braking and during the initial stages of the impact as unbelted occupants, and stand to get more benefit from depowering than unbelted occupants. Arm injuries for drivers are discussed later in this chapter.

Potential Disbenefits of 20-35 Percent Depowering on Fatalities

Table I-5 provides an estimate of the number of fatalities and lives saved by air bags. In this section an estimate is made of what effect depowering has on these lives saved. While there were a number of tests done, the closest to a real world crash are the full vehicle barrier tests. The two test series done on Vehicle I-96 give an indication, for unbelted and belted occupants, what impact 20-35 percent depowering has on injury measurements. These results are shown in Tables III-11 and III-12. It is assumed that the disbenefits of depowering accrue to the entire fleet under the generic sled alternative and to 31 percent of the fleet under the 80 g's alternative. Two different analyses were performed to estimate the effects of depowering in this chapter. One includes only fatalities, while the second methodology includes fatalities and serious injuries. In Chapter V, another analysis focuses on real world experiences with depowered bags.

Method 1 The effect of depowering on fatalities

The first analysis of the impacts of depowering on adult fatalities utilizes results from a NHTSA technical report entitled "Correlation of NCAP Performance with Fatality Risk in Actual Head-On Collisions", January 1994, DOT HS 808-061. This evaluation examined head-on crashes and found that when chest g's are used to partition cars into lower level and poor chest g's groups, the cars with the high chest g's almost always had significantly more fatalities than the cars with lower chest g's. HIC also increases for the vehicles with higher chest g's and the increase in fatalities is a result of both chest g's and HIC increases. Page 90 of this report provides 7 examples of good and poor chest g's and related HIC increases and the percent fatality reduction for the good cars. When chest g's go down by 1 g, the range of results indicate that the overall fatality rate goes down by 0.82 to 1.66 percent. For example, from page 90, a reduction in chest g's from 66.1 to 46.0 resulted in a reduction of the overall fatality rate of 24.2 percent. Thus, the overall fatality rate reduction per 1 g is estimated to be 1.2 [24.2/(66.1-46.0)]. Similarly, Page 87 of the report provides examples of good and poor chest g's and related HIC increases. Examples 4 and 5 on Page 87 were not included in this analysis, since they were not statistically significant, and the last three examples were not included since HIC was lower while chest g's rose, which is not like the test results for depowering. The following discussion focuses on chest g's, rather than HIC, since chest g's had a stronger relationship to fatality risk than HIC and the increases in HIC due to depowering were not that significant and were within the ranges of increases in HIC used in the NCAP analysis.

These results were turned around to analyze the impact of increasing chest g's with depowering rather than lowering chest g's. For example, when an average 66.1 chest g's response was reduced to 46.0 chest g's response, the fatality rate was reduced by 24.2 percent. Turning this around (1/(1-.242) results in a 31.9 percent increase in fatalities if chest g's rise from 46.0 to 66.1. Thus, for every 1 g increase in chest g's, the overall fatality rate of those vehicles is 1.59 percent higher [31.9/(66.1-46)]. The range of answers from the Page 87 and 90 examples is that a 1 g increase in chest g's results in an overall fatality rate increase of 0.96 to 2.80 percent.

1) For unbelted drivers

Depowering by 20-35 percent had very little impact on unbelted driver chest g's in both 30 mph and 35 mph crash testing. In Model I-96, chest g's went up by 2 (49 to 51) in 30 mph testing and went down by 2.9 (61.6 to 58.7) in 35 mph testing. Modelling efforts by Ford show that chest g's should go up by about 2-3 g's with depowering in both instances. NHTSA modelling shows chest g's going up at 40.2 and 64.4 kph, but down at 56.3 kph (see Table III-13). It is believed

that the energy absorbing steering column is the reason that chest g's do not increase much on the driver side with the driver unbelted. Model I-96's test results are very similar to an average of the fleet in NHTSA compliance testing for unbelted drivers of 48 chest g's.

The calculated increase in fatalities using the Model I-96 30 mph test results are:

49 to 51 g's, baseline to depowered tests = a 2 g increase

2 g's x 0.96 to 2.80% = 1.92 to 5.6%

This is applied to .66, which is the fatality rate in purely frontal impacts assuming 1.00 is the fatality rate of baseline vehicles and air bags are 34 percent effective in reducing fatalities for unbelted drivers in pure frontal impacts, which is like the barrier test.

0.66 x 1.0192 to 1.056 = .673 to .697

The increase is (.673 to .697) - .66 = .013 to .037

This .013 to .037 increase in fatalities is applied to the estimated number of fatalities in the 31 to 40 mph unbelted driver cell from Table I-5 of 2,219 resulting in an estimated increase in overall fatalities of

2,219 x .013 to .037 = 29 to 82 fatalities.

The calculated decrease in fatalities using the Model I-96 35 mph test results are:

61.6 to 58.7 g's, baseline to depowered tests = a 2.9 g decrease

2.9 g's x 0.82 to 1.66% = 2.378 to 4.814%

0.66 x [1-(.02378 to .04814)] = .6443 to .6282

The decrease is .66 - (.6443 to .6282) = .0157 to .0318

This .0157 to .0318 decrease in fatalities is applied to the estimated number of fatalities in the 41+ mph unbelted driver cell from Table I-5 of 1,295 resulting in an estimated decrease in overall fatalities of

1,295 x .0157 to .0318 = 20 to 41 fatalities.

Combining the results from Model I-96 testing at 30 and 35 mph, results in an estimated increase in overall unbelted driver fatalities of 9 to 41.

29 to 82

-20 to 41

9 to 41

As discussed in Chapter I, AAMA raised the issue of what percent of the frontal fatalities these estimates should be applied to. The agency believes they apply to at least 34.4 percent of the frontal fatalities and perhaps all 100 percent. Applying them to 34.4 to 100 percent of frontal fatalities results in estimates of 3 to 41 fatalities under the generic sled pulse alternative. Under the assumption that the 80 g's alternative affects 31 percent of the fleet, the impact of depowering would be 1 to 13 fatalities.

2) For belted drivers

The Model I-96 chest g's test results for a belted driver at 35 mph increased significantly from a baseline air bag of 52 chest g's to a depowered air bag of 59.6 chest g's. These results are much higher than modelling would predict and do not make logical sense. Modelling predicts an increase in chest g's from depowering of about 2 g's (see Tables III-14 and III-17). In the same test on the belted passenger dummy, chest g's went down from 52 to 49.6. For this analysis, it is assumed that the modelling results are correct, and that chest g's would increase by 2 g's for belted drivers.

The increase of 2 g's = 2 g's x 0.96 to 2.80% = 1.92 to 5.6%

This is applied to .79, which is the fatality rate in purely frontal impacts assuming 1.00 is the fatality rate of baseline vehicles and air bags are 21 percent effective in reducing fatalities for belted drivers in pure frontal impacts, which is like the barrier test.

0.79 x 1.0192 to 1.056 = .8052 to .834

The increase is (.8052 to .834) - .79 = .0152 to .044

This is applied to the estimated number of fatalities in the 30+ belted driver cells from Table I-5 of 1,650 + 855 = 2,505 resulting in an estimated increase in overall fatalities of

2,505 x .0152 to .044 = 38 to 110 fatalities.

As discussed in Chapter I, AAMA raised the issue of what percent of the frontal fatalities these estimates should be applied to. The agency believes they apply to at least 34.4 percent of the frontal fatalities and perhaps all 100 percent. Applying them to 34.4 to 100 percent of frontal fatalities results in estimates of 13 to 110 fatalities under the generic sled pulse alternative. Under the assumption that the 80 g's alternative affects 31 percent of the fleet, the impact of depowering would be 4 to 34 fatalities.

3) For unbelted passengers

Depowering air bags had a significant impact on chest g's in both 30 mph and 35 mph crash testing. These were the most significant findings of the test program, that unbelted passenger chest g's increased significantly. In Model I-96, chest g's went up by 11 chest g's (48 to 59) in 30 mph testing and went up by 22.6 chest g's (54.5 to 77.1) in 35 mph testing. These large increases were also found in models the agency tested in 30 mph sled tests (I-96 chest g's increased from 47 to 52 with 23 percent depowering, B-94 chest g's increased from 51 to 58 with 30 percent depowering, and D-96 chest g's increased from 42 to 54 with 18 percent depowering). Ford did not present a model on the passenger side. NHTSA modelling efforts (see Table III-15) show that chest g's go up by 6 g's at 25 mph and 10.1 g's at 35 mph with depowering. The vehicle test results from Model I-96 at 35 mph showed a significantly higher impact of depowering than NHTSA modelling.

The calculated impacts of a rise in unbelted passenger chest g's are:

48 to 59 g's, baseline to depowered tests = an 11 g increase

11 g's x 0.96 to 2.80% = 10.56 to 30.8%

This is applied to .66, which is the fatality rate in purely frontal impacts assuming 1.00 is the fatality rate of baseline vehicles and air bags are 34 percent effective in reducing fatalities for unbelted passengers in pure frontal impacts, which is like the barrier test.

0.66 x 1.1056 to 1.308 = .730 to .863

The increase is (.730 to .863) - .66 = .07 to .203

This .07 to .203 increase in fatalities is applied to the estimated number of fatalities in the 31 to 40 mph unbelted passenger cell from Table I-5 of 569 resulting in an estimated increase in overall fatalities of

569 x .07 to .203 = 40 to 116 fatalities.

The calculated increase in fatalities using the Model I-96 35 mph test results are:

54.5 to 77.1 g's, baseline to depowered tests = a 22.6 g decrease

22.6 g's x 0.96 to 2.80% = 21.7 to 63.3%

0.66 x (1.217 to 1.633) = .803 to 1.078

The increase is (.803 to 1.078) - .66 = .143 to .418

This .143 to .418 increase in fatalities is applied to the estimated number of fatalities in the 40+ mph unbelted driver cell from Table I-5 of 393 resulting in an estimated increase in overall fatalities of 393 x .143 to .418 = 56 to 164 fatalities.

Combining the results from Model I-96 testing at 30 and 35 mph, results in an estimated increase in overall unbelted passenger fatalities of 143 to 280.

40 to 116

+56 to 164

96 to 280

As discussed in Chapter I, AAMA raised the issue of what percent of the frontal fatalities these estimates should be applied to. The agency believes they apply to at least 34.4 percent of the frontal fatalities and perhaps all 100 percent. Applying them to 34.4 to 100 percent of frontal fatalities results in estimates of 33 to 280 fatalities under the generic sled pulse alternative. Under the assumption that the 80 g's alternative affects 31 percent of the fleet, the impact of depowering would be 10 to 87 fatalities.

4) For belted passengers

Depowering air bags for the belted passenger resulted in a decrease in chest g's from 52 to 49.6 chest g's in the 35 mph test. Modelling showed almost no change. It is believed chest g's could decrease in this case since the air bag does not hit the chest as hard and lets the belt take up more of the loads. The calculated decrease in fatalities using the Model I-96 35 mph test results are:

52 to 49.6 g's, baseline to depowered tests = a 2.4 g decrease

2.4 g's x 0.82 to 1.66% = 1.968 to 3.984%

0.66 x [1-(.01968 to .03984)] = .647 to .6337

The decrease is .66 - (.647 to .6337) = .013 to .0263

This .013 to .0263 decrease in fatalities is applied to the estimated number of fatalities in the 30+ mph belted passenger cells from Table I-5 of 546 + 285 = 831 resulting in an estimated decrease in overall fatalities of

831 x .013 to .0263 = 11 to 22 fatalities.

As discussed in Chapter I, AAMA raised the issue of what percent of the frontal fatalities these estimates should be applied to. The agency believes they apply to at least 34.4 percent of the frontal fatalities and perhaps all 100 percent. Applying them to 34.4 to 100 percent of frontal fatalities results in estimates of 4 to 22 fatalities under the generic sled pulse alternative. Under the assumption that the 80 g's alternative affects 31 percent of the fleet, the impact of depowering would be 1 to 7 fatalities.

In summation, the potential impacts of depowering air bags using the first method of analysis is shown in Table IV-2.

Table IV-2

Summation of Results
Method 1 Analysis
Estimated Impact of Depowering on Adult Fatalities

80 g's Alternative 20-35 Percent Depowering Impact Affecting 31% of the Fleet Generic Sled Test Alternative, 20-35 Percent Depowering Affecting the Entire Fleet
Unbelted Drivers 1 to 13 3 to 41
Belted Drivers 4 to 34 13 to 110
Driver Total 5 to 47 16 to 151
Unbelted Passengers 10 to 87 33 to 280
Belted Passengers -1 to -7 -4 to -22
Passenger Total 9 to 80 29 to 258
Driver and Passenger Total 14 to 127 45 to 409

Method 2 The Effect of Depowering on Chest Injury and Fatalities

This is a second analysis by the agency using a totally different approach than used in the previous analysis, which relied on the relationship of chest g's in NCAP testing to FARS data, to estimate the effects of depowering air bags. This analysis includes an estimate of the effect depowering has on fatalities and on AIS 3, AIS 4, and AIS 5 non-fatal injuries.

The analysis is based on NASS-CDS data and the cadaver test data provided by NHTSA's Biomechanics Research Division. The NASS-CDS crash data for 1991-1995 were utilized and the population of occupants was restricted to front seat occupants ages 16 to 76. The method is illustrated below using the numbers pertaining to the sub-population of occupants with air bag deployed during the crash. The calculation shown assumes a 33% increase in chest acceleration and provides estimates of the corresponding increase in chest injuries and fatalities. The analysis assumes that depowering causes a given proportionate increase in occupant chest g's for all crash involved individuals regardless of their crash delta V and, through the following procedure, estimates the additional fatalities expected as a result. The same method can be applied to other changes in chest acceleration and to other population (e.g. belted or unbelted occupants) to obtain the corresponding changes in injuries and fatalities. The results of such calculations are shown in the last section.

The first step in the analysis is to estimate chest injury numbers and chest injury rates at AIS 3+ (AIS 3 or greater), AIS 4+, and AIS 5+ levels among individuals in crashes within specified delta-v ranges, which for this study were chosen to be 0-10, 11-20, 21-30, 31-40, and 41 and up. The results are presented in Table IV-3.

Table IV-3

Injury Rates (front seat occupants, age 16 to 76, air bag deployed)

v Number of Individuals AIS 3 + AIS 4 + AIS 5 +
Number Rate Number Rate Number Rate
0-10 85,614 0 0.00000 0 0.00000 0 0.00000
11-20 196,451 655 0.00333 72 0.00037 11 0.00006
21-30 36,036 900 0.02498 457 0.01268 107 0.00297
31-40 3,033 715 0.23574 552 0.18200 101 0.03330
41 and up 790 209 0.26456 145 0.18331 145 0.18331

These raw injury rates were then used to estimate a hypothetical injury curve for each level of injury (AIS 3+, AIS 4+, AIS 5+) based on the probit model. That is, it is assumed that the probability of injury AIS 3 or greater at a given delta-v is P(v)=(a+bv), where

is the cumulative normal distribution function and a and b are coefficients estimated using the maximum likelihood method. The same model was used to estimate the injury rate curves for AIS 4+ and AIS 5+ injuries. The calculation was done using the SAS statistical software. The resulting estimated probabilities are presented in Table IV-4.

Table IV-4

Estimated Injury Probabilities

v AIS 3 + AIS 4 + AIS 5 +
Probit Estimate Raw Rate Probit Estimate Raw Rate Probit Estimate Raw Rate
0-10 0.00016 0.00000 0.00001 0.00000 0.00000 0.00000
11-20 0.00306 0.00333 0.00058 0.00037 0.00009 0.00006
21-30 0.02948 0.02498 0.01253 0.01268 0.00262 0.00297
31-40 0.15029 0.23574 0.10899 0.18200 0.03353 0.0330
41 and up 0.42782 0.26456 0.41166 0.18331 0.19198 0.18331

Exhibit 1 shows graphs of the original data points and the smoothed curves.

Exhibit 1

The next step in the analysis is to obtain the modified probabilities of chest injuries at the AIS 3+, AIS 4+, and AIS 5+ levels to reflect the change in the injury probability due to the change in chest acceleration. This is done utilizing the injury curves obtained from the cadaver test data, cadaver testing with air bags and no belts (see Figure II-4 and Table II-7). These curves are shown in Exhibit 2.

The cadaver injury risk curves give a probability of injury at a given AIS level (AIS 3+, AIS 4+, AIS 5+) as a function of chest acceleration measured in g units. It is then possible to use these curves to assess the change in probability of injury corresponding to a change in the chest acceleration. For example, at 60 g the probability of an AIS 3+ injury is about 0.14553 (according to the cadaver injury risk curve), and if the acceleration is increased 33.33% to 80 g, the probability of an AIS 3+ chest injury increases to 0.35429.

On the other hand, the injury risk curves obtained from crash data (as presented in Exhibit 1), give the probability of injury at a given delta-v level. In order to find how the probability of an injury on the crash injury curve should change when the chest acceleration changes, the increase in chest acceleration is translated to the corresponding increase in delta-v. For each point on an crash injury risk curve, i.e., for each probability of injury on the curve corresponding to a specific delta-v, we find what is the chest acceleration, in g units, that produces that same probability of injury on the cadaver injury curve. Then the modified probability of the injury is determined, corresponding to an increase in chest acceleration by 33.33%. The new probability

Exhibit 2

of injury is then mapped back to the crash injury risk curve and the modified delta-v level corresponding to that new probability is obtained. When this is done for every point on the crash injury rate curve, we obtain a translation of the original delta-v levels to modified delta-v levels corresponding to a 33.33% increase in chest acceleration. The resulting modified delta-v for the mid-points of the delta-v ranges used above for the three injury risk curves (AIS 3+, AIS 4+, AIS 5+) are shown in Table IV-5.

Table IV-5

Delta-v levels corresponding to increase in injury
due to a 33.33% increase in chest acceleration

Original v Modified v
AIS 3+ 		Modified v
AIS 4+ 		Modified v
AIS 5+
5 5.0000 5.0000 5.0000
15 15.0000 15.0000 15.0000
25 28.2649 27.4085 28.2795
35 42.7516 42.2126 43.0106
45 57.8167 57.1073 over 60

Ideally, these delta-v should be the same for all three curves. In reality, there are some differences. Based on judgment and experience with the cadaver tests, it was decided that the shift corresponding to the AIS 4+ injury risk curve is the most appropriate to use. Consequently, modified crash injury risk curves were constructed using the modified delta-v values for AIS 4+ injuries from Table IV-9 in the equation P(v)=(a+bv). The modified probabilities of AIS 3+, AIS 4+, AIS 5+ are presented in Table IV-6.

Table IV-6

Modified Injury Probabilities
(assuming 33.33% increase in chest acceleration)

v AIS 3 + AIS 4 + AIS 5 +
Estimated Probability Modified Probability Estimated Probability Modified Probability Estimated Probability Modified Probability
0-10 0.00016 0.00016 0.00001 0.00001 0.00000 0.00000
11-20 0.00306 0.00306 0.00058 0.00058 0.00009 0.00009
21-30 0.02948 0.04619 0.01253 0.02288 0.00262 0.00523
31-40 0.15029 0.33733 0.10899 0.30698 0.03353 0.12748
41 and up 0.42782 0.80266 0.41166 0.84084 0.19198 0.61502

The modified injury risk curves are shown in Exhibit 3.

Using the original injury curves and the modified injury risk curves, it is now possible to estimate the change in the number of chest injuries (at AIS 3+, AIS 4+, and AIS 5+ levels) due to the change in chest acceleration by 33.33%. For each delta-v category, the number of individuals with a given level of injury is estimated using the original injury risk curve by multiplying the total number of individuals in that delta-v category by the estimated probability of injury. The modified number of chest injuries is obtained by multiplying the same total by the modified

Exhibit 3

probability. The difference between the two numbers represents the change in the number of injuries at the given level due to the change in chest acceleration. The results are presented in Table IV-7.

Table IV-7

Estimated numbers of injuries and modified numbers of injuries
(assuming 33.33% increase in chest acceleration)

v AIS 3 + AIS 4 + AIS 5 +
Estimated Number Modified Number Estimated Number Modified Number Estimated Number Modified Number
0-10 14 14 1 1 0 0
11-21 600 600 114 114 17 17
21-30 1062 1665 451 824 94 188
31-40 456 1023 331 931 102 387
41 and up 338 634 326 665 152 486

The estimated and modified numbers of injuries at AIS 3, AIS 4, and AIS 5 levels are obtained by subtracting the number of AIS 4+ injuries from the number of AIS 3+ injuries, and the number of AIS 5+ injuries from the number of AIS 4+ injuries, and assuming that AIS 5+ injuries represent the number of AIS 5 injuries (since the number of AIS 6 injuries is relatively small). In the case where, due to error in statistical estimation, the number of AIS 4+ injuries is estimated as greater than the number of AIS 3+ injuries, the result is assumed to be zero. These numbers are presented in Table IV-8.

Table IV-8

Estimated numbers of chest injuries and modified numbers of chest injuries
(assuming 33.33% increase in chest acceleration).

v AIS 3 AIS 4 AIS 5
Estimated Number Modified Number Estimated Number Modified Number Estimated Number Modified Number
0-10 13 13 1 1 0 0
11-21 486 486 97 97 17 17
21-30 611 841 357 636 94 188
31-40 125 92 229 544 102 387
41 and up 12 0 174 179 152 486

By summing the estimated and modified numbers of injuries across delta-v categories the total increases in AIS 3, AIS 4, and AIS 5 injuries are calculated. In this example, the percent changes are: AIS 3 14.84%, AIS 4 69.81%, AIS 5 195.34%. The increase in AIS 3+ injuries is 59.35%.

In order to estimate the change in the number of fatalities due to increase in chest acceleration, the following method is used. It is assumed that the number of fatalities in a given population of crash-involved individuals can be predicted based on their two highest AIS injuries. The fatality rates are obtained for subpopulations with highest two AIS injuries (5,5), (5,4), (5,3), (5,2), (5,1), (5,0), (4,4), (4,3), . . . , (1,0), (0,0). The number of fatalities in a given population is then estimated as the weighted average of the number of individuals in these categories with weights equal to the fatality rates. This would give the exact number of fatalities in the given population if the fatality rates used as the weights were based on the fatalities in this same population. However, in the present analysis, fatality rates in the above injury categories are calculated using the largest population available (all individuals in 1991-1995 NASS-CDS age 16 to 76 with known injury levels). Since these rates are applied to the population of front-seat occupants with air bag deployed, the predicted number of fatalities already differs from the actual number at this preliminary step of the analysis. However, it was concluded that these fatality rates were closer to the actual, hypothetical, fatality rates compared to fatality rates obtained using smaller sub-populations, and hence they would provide a better basis for estimating the change in fatalities due to changes in chest acceleration.

Furthermore, the raw fatality rates were smoothed. The smoothing was done in groups determined by the highest AIS, that is (0,0) was treated as one group, (1,0),(1,1) was the second group, (2,0),(2,1),(2,2) was the third group, (3,0),(3,1),(3,2),(3,3) was the fourth group, (4,0),(4,1),(4,2),(4,3),(4,4) was the fifth group, and (5,0),(5,1),(5,2),(5,3),(5,4),(5,5) was the last group. It was observed that within each of these groups, the raw fatality rates were not monotone, and it was concluded that smoothed fatality rates monotone within each group would be more realistic. For each of the groups, a probit curve was fit, using the same procedure as that utilized to smooth the injury curves above. In addition, since fatality rates corresponding to highest two AIS injuries (2,0), (3,0), and (4,0) appear to be outliers (they are much higher than the fatality rates in the categories (2,1), (3,1), (4,1), respectively) and are based on small numbers of observations, they were adjusted by combining these categories, as well as categories (1,0) and (1,1) for consistency. The raw fatality rates and the smoothed fatality rates used in the analysis are presented in Table IV-9. Figure IV-1 shows the graphs of the original and the smoothed fatality rates.

Table IV-9

Fatality Rates (original and smoothed)

Two highest AIS Raw Fatality Rate Smoothed Fatality Rate
(0,0) 0.00000 0.00000
(1,0) 0.00016 0.00016
(1,1) 0.00026 0.00026
(2,0) 0.01172 0.00175
(2,1) 0.00164 0.00454
(2,2) 0.01592 0.01075
(3,0) 0.01654 0.01129
(3,1) 0.01969 0.02317
(3,2) 0.04566 0.04417
(3,3) 0.07819 0.07837
(4,0) 0.16526 0.08770
(4,1) 0.10738 0.15317
(4,2) 0.38647 0.24484
(4,3) 0.27511 0.35991
(4,4) 0.59530 0.48940
(5,0) 0.50602 0.19180
(5,1) 0.23202 0.27100
(5,2) 0.56063 0.36380
(5,3) 0.34301 0.46540
(5,4) 0.71548 0.56932
(5,5) 0.59504 0.66863

Figure IV-1

The actual and predicted numbers of fatalities in the sub-population of front-seat occupants with air bag deployed are shown in Table IV-10.

Table IV-10

Fatality numbers: actual and predicted from fatality rates

v Actual number of fatalities Predicted number of fatalities based on fatality rates
0-10 0 55
11-20 165 397
21-30 560 606
31-40 722 489
41 and up 164 103
Total 1612 1650

In order to predict the number of fatalities when chest acceleration is increased, a procedure is used to replace the chest injuries in the data with 'synthetic', hypothetical chest injuries, based on the probability of a given AIS level chest injury corresponding to the delta-v of the crash .

First, the probabilities of AIS 3, AIS 4 and AIS 5 injuries (estimated and modified due to the 33.33% increase in chest acceleration) in the delta-v categories as above are obtained from the probabilities of AIS 3+, AIS 4+, and AIS 5+ injuries (estimated and modified, as in Table IV-6). The results are presented in Table IV-11 (again, negative values are replaced by zeroes).

Table IV-11

Probabilities of injuries and modified probabilities of injuries
(assuming 33.33% increase in chest acceleration).

v AIS 3 AIS 4 AIS 5
Estimated Probability Modified Probability Estimated Probability Modified Probability Estimated Probability Modified Probability
0-10 0.000152 0.000152 0.00001 0.00001 0.00000 0.00000
11-20 0.002477 0.002477 0.00049 0.00049 0.00009 0.00009
21-30 0.016956 0.023316 0.00991 0.01765 0.00262 0.00523
31-40 0.041305 0.030354 0.07545 0.17950 0.03353 0.12748
41 and up 0.016167 0.000000 0.21968 0.22582 0.119198 0.61502

The injury curves for AIS 3, AIS 4, and AIS 5 injuries are shown in Exhibit 4.

Exhibit 4

The procedure of replacing the actual injuries by the 'synthetic' injuries consists of removing all actual chest injuries in the dataset and replacing each individual in the population with four individuals with an additional AIS 3, AIS 4, AIS 5, and AIS 0 injury, respectively. These individuals have an attached weighting factor equal to the corresponding probability of their AIS 3, AIS 4, AIS 5, and AIS 0 chest injuries for the delta-v of the crash they were involved in. It is assumed that the probability of an AIS 0 injury equals one minus the sum of the probabilities of AIS 3, AIS 4, and AIS 5 injuries. In this way, a new population quadrupled in size is constructed. The number of fatalities in this new population is predicted using the method based on their two highest AIS injuries as above. Although the new population is quadrupled in size, the application of the weighting factors assures that the predicted number is compatible with the original number. The differences are due to the fact that instead of the actual injury rates, the injury rates from the 'smoothed' curves are used. The results are presented in Table IV-12.

Table IV-12

Fatality numbers predicted from fatality rates and fatality numbers predicted
using the imputed chest injuries

v Predicted number of fatalities based on fatality rates Predicted number of fatalities using the imputed chest injuries
0-10 55 53
11-21 397 384
21-30 606 524
31-40 489 483
41 and up 103 136
Total 1650 1581

The same procedure of estimating the number of fatalities in the population can now be used with the modified injury probabilities serving as the weighting factors. This produces the predicted number of injuries under the assumption that chest acceleration increased 33.33%. The results are presented in Table IV-13.

Table IV-13

Fatality numbers predicted using the imputed chest injuries using the original conditions and assuming a 33.33% increase in chest acceleration

v Predicted number of fatalities under the original conditions Predicted number of fatalities assuming 33.33% increase in chest acceleration
0-10 53 53
11-21 384 384
21-30 524 608
31-40 483 638
41 and up 136 257
Total 1581 1941

Thus, the estimated increase in chest injury-related fatalities is 22.78%.

The following table presents the results of these calculations for the following chest acceleration increases: 4.167% (48g to 50g), 14.6% (52g to 60g), 22.9% (48g to 59g), 33.33% (60g to 80g), 41.5% (54.5g to 77.1g).

Table IV-14

Increases/decreases in injuries and fatalities due to chest acceleration increase/decrease. Air bag deployed, age between 16 and 76, NASS-CDS 1991-1995
(population 35% unbelted, 65% belted)

Chest acceleration AIS 3+ AIS 3 AIS 4 AIS 5 Fatal
g 4.167% 5.67% 1.60% 7.46% 15.34% 2.09%
g 14.6% 21.74% 3.93% 27.86% 68.22% 8.54%
g 22.9% 36.64% 6.09% 45.22% 120.82% 14.49%
g 33.33% 59.35% 12.35% 69.81% 195.34% 22.78%
g 41.5% 76.52% 14.11% 87.06% 264.93% 29.94%
g's go down 4.62 % -10.85% -4.73% -14.69% -22.74% -3.48%

The target population for the second method of analyzing the impacts of depowering has been examined for injuries considering only those frontal impacts in which the chest injury has the maximum AIS level. For the injury part of the analysis, only chest injuries are considered. Thus, the percent changes in injuries are applied only to chest injuries. However, the fatality percentages take into account all fatalities, whether chest injury is the maximum AIS injury or not, and the percent increase in fatalities applies to all fatalities, not just the 50 percent that are related to chest injury. Unlike the NCAP analysis, where the increased fatality rate is only applied to those cases of delta V of 31 mph or more, this analysis applies to all delta V's. The percent increases were calculated assuming a percentage increase in chest g's at all delta V's and determining the impact on fatalities.

1) For unbelted drivers

Depowering air bags is assumed to result in an increase in chest g's of 2 g's. Since the average compliance test score with air bags is 48 g's, it was assumed a 2 g's increase over 48 g's (48 to 50) results in a 4.167 percent increase in chest g's. As shown Table IV-14, the analysis estimated that a 4.167 percent increase in chest g's results in a 2.09 percent increase in fatalities. Thus, the estimated impacts of depowering on the unbelted driver are:

8,939 fatalities - 1,686 saved by air bags (see Table I-5) = 7,253 - [.7 catastrophic crashes in the 41+ mph delta V cell before air bags x (1295)] =

7,253 - 907 = 6,346 x .0209 = 133 fatality increase

As discussed in Chapter I, AAMA raised the issue of what percent of the frontal fatalities these estimates should be applied to. The agency believes they apply to at least 34.4 percent of the frontal fatalities and perhaps all 100 percent. Applying them to 34.4 to 100 percent of frontal fatalities results in estimates of 46 to 133 under the generic sled pulse alternative. Under the assumption that the 80 g's alternative affects 31 percent of the fleet, the impact of depowering would be 14 to 41 fatalities.

2) For belted drivers

Depowering is assumed to increase chest g's by 2 or 4.167 percent. As shown Table IV-14, the analysis estimated that a 4.167 percent increase in chest g's results in a 2.09 percent increase in fatalities. Thus, the estimated impacts of depowering on the unbelted driver are:

4,737 fatalities (see Table I-5) - 616 saved by air bags (see Table I-5) = 4,121 - (.7 catastrophic crashes in the 41+ mph delta V cell x 855) =

4,121 - 599 = 3,522 x .0209 = 74 fatality increase

As discussed in Chapter I, AAMA raised the issue of what percent of the frontal fatalities these estimates should be applied to. The agency believes they apply to at least 34.4 percent of the frontal fatalities and perhaps all 100 percent. Applying them to 34.4 to 100 percent of frontal fatalities results in estimates of 25 to 74 under the generic sled pulse alternative. Under the assumption that the 80 g's alternative affects 31 percent of the fleet, the impact of depowering would be 8 to 23 fatalities.

3) For unbelted passengers

Depowering is assumed to increase chest g's in the 31-40 mph delta V range by 48 to 59 or 22.9 percent and in the 41+ mph delta V range by 54.5 to 77.1 or a 41.5 percent increase. As shown in Table IV-14, the analysis estimated that a 22.9 percent increase in chest g's results in a 14.49 percent increase in fatalities (applied to the 0-40 mph delta V cases) and a 41.5 percent increase in chest g's results in a 29.94 percent increase in fatalities (applied to the 41+ mph cases). Thus, the estimated impacts of depowering on the unbelted passenger are:

0-40 mph

1,892 fatalities (see Table I-5) - 421 saved by air bags (see Table I-5) = 1,471

1,471 x 0.1449 = 213

41+ mph

393 fatalities - 70 saved by air bags = 323 - (.7 catastrophic crashes in the 41+ mph delta V cell x 393) = 323 - 275 = 48 x 0.2994 = 14 fatality increase

The total increase in fatalities for depowering for unbelted passengers = 227

As discussed in Chapter I, AAMA raised the issue of what percent of the frontal fatalities these estimates should be applied to. The agency believes they apply to at least 34.4 percent of the frontal fatalities and perhaps all 100 percent. Applying them to 34.4 to 100 percent of frontal fatalities results in estimates of 78 to 227 under the generic sled pulse alternative. Under the assumption that the 80 g's alternative affects 31 percent of the fleet, the impact of depowering would be 24 to 70 fatalities.

4) For belted passengers

Depowering decreased chest g's in the belted passenger test from 52 to 49.6 a decrease of 4.62 percent. As shown in Table IV-14, this would result in a decrease in fatalities of 3.48 percent. Thus, the estimated impacts of depowering on the unbelted passenger are:

1,567 fatalities - 223 saved by air bags = 1,344 - (.7 catastrophic crashes in the 41+ mph delta V cell x 285) = 1,344 - 200 = 1,144 x 0.0348 = 40 fatality decrease

As discussed in Chapter I, AAMA raised the issue of what percent of the frontal fatalities these estimates should be applied to. The agency believes they apply to at least 34.4 percent of the frontal fatalities and perhaps all 100 percent. Applying them to 34.4 to 100 percent of frontal fatalities results in estimates of a decrease in fatalities of 14 to 40 under the generic sled pulse alternative. Under the assumption that the 80 g's alternative affects 31 percent of the fleet, the impact of depowering would be a decrease in fatalities of 4 to 12 fatalities.

The total impacts of depowering on fatalities in this analysis are shown in Table IV-15. Table IV-16 shows the range of estimated impact. Method 1 had the lowest and highest estimates in the range.

Table IV-15

Summation of Results
Method 2 Analysis
Estimated Impact of Depowering on Adult Fatalities

80 g's Alternative
20-35 Percent Depowering Affecting 31% of the Fleet 		Generic Sled Test Alternative, 20 to 35 Percent Depowering Affecting the Entire Fleet
Unbelted Drivers 14 to 41 46 to 133
Belted Drivers 8 to 23 25 to 74
Driver Total 22 to 64 71 to 207
Unbelted Passengers 24 to 70 78 to 227
Belted Passengers -4 to -12 -14 to - 40
Passenger Total 20 to 58 64 to 187
Driver and Passenger Total 42 to 122 135 to 394

Table IV-16

Summation of Results
Method 1 and 2 Analyses
Estimated Impact of Depowering on Adult Fatalities

80 g's Alternative
20-35 Percent Depowering Impact Affecting 31% of the Fleet 		Generic Sled Test Alternative, 20-35 Percent Depowering Affecting the Entire Fleet
Unbelted Drivers 1 to 13* 3 to 41*
Belted Drivers 4 to 34 13 to 110
Driver Total 5 to 47* 16 to 151*
Unbelted Passengers 11 to 87 34 to 280
Belted Passengers -1 to -7* -4 to -22
Passenger Total 10 to 80 30 to 258
Driver and Passenger Total 15 to 127 46 to 409

* The high end of the range is taken from the methodology that derived the highest total of 127 or 409. Individual estimates in those cells marked with an * were higher than noted here, but that methodology produced a lower total estimate than 409.

Note: In Attachment 3, page 21 of the AAMA docket submission (74-14-N108-112), AAMA estimated the disbenefits of depowering based on information provided in the PRE to be 12 to 120 lives per year assuming the full fleet has depowered air bags.

The Impacts of Depowering on Chest Injuries

The same analysis can be performed for AIS 3, AIS 4 and AIS 5 injuries utilizing the estimated increase and decrease in injuries from Table IV-14.

An estimate was made based on 1991-95 NASS data of the annual number of maximum chest injuries in frontal impacts by AIS level. These estimates have not been adjusted to take into account how air bags could affect the injuries. It is not believed that the annual number would change significantly after adjusting for air bags. This analysis estimates the effect depowering would have on AIS level injuries. These are shown in Table IV-17 for driver and passenger side.

Table IV-17

Annual Estimate of the Number of

Maximum Chest AIS 3-5 Injuries

AIS Level Driver Passenger
3 11,895 2,486
4 2,900 829
5 960 313

1) For drivers (unbelted and belted)

Depowering air bags is assumed to result in an increase in chest g's of 2 g's for both belted and unbelted drivers. Since the average compliance test score with air bags is 48 g's, it was assumed a 2 g's increase over 48 g's (48 to 50) results in a 4.167 percent increase in chest g's. As shown in the Table IV-14, the analysis estimated that a 4.167 percent increase in chest g's results in a 1.60 percent increase in AIS 3, a 7.46 percent increase in AIS 4 and a 15.34 percent increase in AIS 5 injuries. Thus, the estimated impacts of depowering on drivers are:

As discussed in Chapter I, AAMA raised the issue of what percent of the frontal fatalities these estimates should be applied to. The agency believes they apply to at least 30.9 percent of the frontal AIS 3-5 injuries and perhaps all 100 percent. Applying them to 30.9 to 100 percent of frontal serious injuries results in estimates of 171 to 553 under the generic sled pulse alternative. Under the assumption that the 80 g's alternative affects 31 percent of the fleet, the impact of depowering would be 53 to 171 AIS 3-5 injuries.

2) For unbelted passengers

Depowering is assumed to increase chest g's in the 31-40 mph delta V range by 48 to 59 or 22.9 percent and in the 41+ mph delta V range by 54.5 to 77.1 or a 41.5 percent increase. As shown in Table IV-14, the analysis estimated that a 22.9 percent increase in chest g's results in a 6.09 percent increase in AIS 3, a 45.22 percent increase in AIS 4 and a 120.82 percent increase in AIS 5 injuries. A 41.5 percent increase in chest g's results in a 12.35 percent increase in AIS 3, a 87.06 percent increase in AIS 4 and a 264.93 percent increase in AIS 5 injuries. Averaging these two estimates for these severe injuries results in the following estimated impacts of depowering on unbelted passengers:

As discussed in Chapter I, AAMA raised the issue of what percent of the frontal fatalities these estimates should be applied to. The agency believes they apply to at least 30.9 percent of the frontal AIS 3-5 injuries and perhaps all 100 percent. Applying them to 30.9 to 100 percent of frontal serious injuries results in estimates of 152 to 491 under the generic sled pulse alternative. Under the assumption that the 80 g's alternative affects 31 percent of the fleet, the impact of depowering would be 47 to 152 AIS 3-5 injuries.

3) For belted passengers

Depowering decreased chest g's in the belted passenger test from 52 to 49.6 a decrease of 4.62 percent. As shown in Table IV-14, this would result in a decrease in injuries of 4.73 percent in AIS 3, a 14.69 percent decrease in AIS 4 and a 22.74 percent decrease in AIS 5 injuries. Thus, the estimated impacts of depowering on the unbelted passenger are:

As discussed in Chapter I, AAMA raised the issue of what percent of the frontal fatalities these estimates should be applied to. The agency believes they apply to at least 30.9 percent of the frontal AIS 3-5 injuries and perhaps all 100 percent. Applying them to 30.9 to 100 percent of frontal serious injuries results in estimates of 62 to 202 under the generic sled pulse alternative. Under the assumption that the 80 g's alternative affects 31 percent of the fleet, the impact of depowering would be 19 to 63 AIS 3-5 injuries.

Table IV-18 summarizes the estimated impact of depowering on AIS 3-5 injuries.

Table IV-18

Summation of AIS 3-5 Injury Results

Drivers 80 g's Belted 80 g's Unbelted 80 g's Total Generic Sled Belted Generic Sled Unbelted Generic Sled Total
AIS 3 12 to 38 6 to 21 18 to 59 38 to 123 21 to 67 59 to 190
AIS 4 13 to 43 8 to 24 21 to 67 43 to 140 23 to 76 66 to 216
AIS 5 9 to 30 5 to 16 14 to 46 30 to 96 16 to 51 46 to 147
Total 34 to 111 19 to 61 53 to 172 111 to 359 60 to 194 171 to 553
R. F. Pass.
AIS 3 -7 to -24 8 to 27 1 to 3 -24 to -77 27 to 88 3 to 11
AIS 4 -8 to -25 19 to 60 11 to 35 -24 to -79 60 to 192 36 to 113
AIS 5 -4 to -14 20 to 65 16 to 51 -14 to -46 65 to 211 51 to 165
Total -19 to -63 47 to 152 28 to 89 -62 to -202 152 to 491 90 to 289
Driver and R.F. Pass.
AIS 3 5 to 14 14 to 48 19 to 62 14 to 46 48 to 155 62 to 201
AIS 4 5 to 18 27 to 84 32 to 102 19 to 61 83 to 268 102 to 329
AIS 5 5 to 16 25 to 81 30 to 97 16 to 50 81 to 262 97 to 312
Total AIS 3-5 15 to 48 66 to 213 81 to 261 49 to 157 212 to 685 261 to 842

Arm Injuries

Table IV-19 presents 1988 to 1995 NASS data on AIS 0-3 arm injuries and air bag interaction. The data represent both male and female drivers in frontal crashes and separates the driver into either air bag deployment or vehicles with no air bag and therefore no air bag deployment.

Table IV-19

Arm Injuries and Air Bag Interaction
Vehicles with Air Bags in Frontal Crashes 1988 to 1995

AIS Levels No. of AIS Injuries Rate of Injury
0 (no injuries) 262,678 ---
1 186,286 0.402
2 8,588 0.019
3 5,674 0.012
Vehicles With No Air Bag In Frontal Crashes 1988 to 1995
AIS Levels No. of AIS Injuries Rate of Injury
0 (no injuries) 7,730,331 ---
1 1,255,548 0.137
2 151,018 0.017
3 33,565 0.004

At AIS 1 and AIS 3 levels of injury, the injury rate was higher for vehicles equipped with air bags. Combining AIS 2 and 3 injuries, the rate of arm injury for air bag equipped vehicles was 3.1 percent, while the rate for vehicles not equipped was 2.1 percent.

Arm injuries were also examined for males and females separately. The rate of arm injuries at the AIS 2 and 3 level are essentially the same for males and females when no air bag is installed in the vehicle. However, based on 1988-95 NASS cases, when an air bag is present, the rate of AIS 2-3 arm injury is much higher for females (4.93 percent) than for males (1.81 percent). Thus, crash data corroborates the biomechanical finding that men's arms have a higher threshold before breaking than women's arms.

Table IV-20 provides estimates of the number of driver arm injuries that would occur in frontal crashes if all vehicles had air bags or if no vehicles had air bags.

Table IV-20

Estimated Annual Fleet AIS levels if all vehicles have a driver air bag

AIS Levels No. Of AIS Injuries
1 281,036
2 16,628
3 8,378
Estimated Annual Fleet AIS levels if no air bag vehicles
AIS Levels No. Of AIS Injuries
1 128,306
2 15,433
3 3,430

Twelve vehicles were tested to determine the aggressiveness of their air bags (see Table IV-21). Some vehicles had changed the driver air bag design in 1996. The vehicles are coded A through L. The configurations V-1 and V-3 are the orientation of the arm device relative to the steering wheel, and the distance between the arm and the steering wheel.

For the V-1 configuration, the dummy hand/elbow is in the "clock" position of 11 - 4, with the nominal distance from the wheel of .5". The wheel rotation is 90o counter clock wise.

For the V-3 configuration, the dummy hand/elbow is in the "clock" position of 10 - 4, with the nominal distance from the wheel of .5". The wheel rotation is 60o counter clock wise.

Table IV-21

Peak Resultant Moments

Vehicle type
(V-1 Configuration) 		1994 1996 1996 Depowered
E 181.6 89.1
B 54.6
F 234.6 173.0
H 239.9
D 147.1
G 212.6
A 192.9 156.8
I 143.9
AVERAGE 203.03 152.1
V-3 configuration
E 185.4 82.8
B 125.7
F 234.7 120.7 85.0 (D-35%)
H 241.0
D 235.6
G 244.2
A 239.1 133.4
I 172.2 167.0 (D-25%)
150.0 (D-42%)
AVERAGE 219.7 169.45

Criteria

From cadaver testing by the agency, estimated ranges for arm injuries have been developed. The ranges are: 70 to 90 N m for females and 120 to 150 N m for males. This range means that for females any number above 90 Nm the individual would experience a broken arm, and any recording below 70 N m the individual should not suffer a broken arm. Similarly for men, above 150 N m the individual should suffer a broken arm, and for numbers below 120 N m the individual should not suffer a broken arm. These results are for research purposes only and should not be taken as rigid limits, since in reality these ranges must overlap. The agency is in the formative stage regarding knowledge about arm injuries.

The data in Table IV-21 indicate that in 1994 all drivers with their arms in these positions would have suffered broken arms because the aggressiveness of the air bags was above the injury threshold for both sexes. In 1996, one vehicle was below the female injury range, and two vehicles were within the female injury range. The data show that in 1996 there were three vehicles below the male injury range, and five vehicles within the male injury range.

For the depowered bags, the F-model bag that was depowered by 35 percent dropped down from the male injury range into the female injury range. For the I-model bag, 25 and 42 percent depowering of the air bag were not enough to lower the aggressiveness of the bag to within the male injury range.

On one vehicle, depowering the air bag by 35 percent resulted in a 29.6 percent reduction in measured aggressiveness, while on the other vehicle, depowering the air bag by 25 percent and 42 percent resulted in 3.02 and 12.9 percent reduction in aggressiveness respectively. Based on these two air bag systems, there does not appear to be any correlation between the percent that different air bags are depowered and the percents reduction in arm injury aggressiveness.

Assuming that the AIS 2-3 arm injury ranges are accurate, that the MY 94 and MY 96 vehicles tested are representative of the fleet, depowering in the 20-30 percent range lowers arm measurements 3 to 30 percent, and that V-1 and V-3 configuration have an equal chance of occurrence, estimates have been made of the number of arm injuries that would result per year if all vehicles had air bags and are shown in Table IV-22.

Table IV-22

Estimated Arm Injuries AIS 2-3

Based on Number of injuries Savings Compared to MY 94 % Savings
MY 94 Vehicles 25,006 ---- ----
MY 96 Vehicles 20,680 4,326 17.3
Depowered 96
Effective 3% 19,880 5,126 20.5
Effective 30% 16,254 8,752 35.0

In summary, if all air bags were depowered under the generic sled alternative an estimated 5,126 to 8,752 AIS 2-3 arm injuries could be reduced. Under the 80 g's alternative, assuming 31 percent of the fleet is depowered, the estimates would be 1,589 to 2,713.

Viewing the peak resultant data one observes that the vehicle coded B in the V-1 configuration for 1996, had a score of 54.6 peak moment. If this is possible for one vehicle, then maybe other vehicle manufacturers can modify their air bags to produce similar scores while meeting the current FMVSS 208 Standards.

Relationship of Countermeasures

There is an interrelationship of benefits from countermeasures being considered by the agency. The three countermeasures under consideration include labelling, manual cut-off switches and depowered air bags.

If labelling were 100 percent effective, there would be no need for depowering air bags or for manual cut-off switches, except in those vehicles that have a short rear seat or no rear seat. However, clearly labelling will not be 100 percent effective. It is very difficult to determine how effective labelling and other consumer information will be in getting people to change their behavior and put children in the rear seat. To the extent that labelling is effective, the potential benefits of depowering is reduced. For example, if labelling results in 10 percent of child occupants being moved into the rear seat, the benefits of depowering will be reduced by 10 percent. Similarly, to the extent that manual cut-off switches are used correctly, the potential benefits of depowering air bags are reduced. However, the combination of labeling and depowering will result in greater total benefits than either approach alone.

Safety Belt Use

In 1995, about 36 percent of fatally injured passenger car and light truck drivers and right front seat passengers were belted. Using our effectiveness estimates, about 52.6 percent were belted in potentially fatal crashes. If 75 percent of driver and right front seat occupants were belted in potentially fatal crashes, an additional 4,000 lives would be saved annually. Similarly, if 80 percent of these occupants were belted in potentially fatal crashes, an additional 4,900 lives would be saved.

If belt use increased, obviously it would be good for safety, it would have the following effects on this notice on depowering:

1) For infants in rear-facing child restraints -- no impact.

2) For forward-facing children -- increased belt use will reduce the number of children killed by current air bags. This will lower the potential benefits of depowering for children. But, overall lives saved will increase.

3) For adults drivers -- increasing belt use will make a small reduction in the at risk population. Based on the Holden bag analysis, increasing belt use could increase the benefits of depowering by increasing the effectiveness of air bags for belted occupants.

4) For passengers over 12 -- Based on the Holden bag analysis, increasing belt use could increase the benefits of depowering by increasing the effectiveness of air bags for belted occupants. For unbelted passengers, increasing belt use would reduce the disbenefits of depowering by lowering the number of unbelted occupants.

The bottom line is that although increasing safety belt use reduces the benefits of depowering for one part of the population, there will be very large increases in the number of people saved by occupant restraints.