DEPARTMENT OF TRANSPORTATION

National Highway Traffic Safety Administration

49 CFR Parts 552, 571, 585, and 595

[Docket No. NHTSA 99-6407; Notice 1]

RIN 2127-AG70

Federal Motor Vehicle Safety Standards;
Occupant Crash Protection

AGENCY: National Highway Traffic Safety Administration (NHTSA), DOT.

ACTION: Supplemental notice of proposed rulemaking (SNPRM).

SUMMARY: In September 1998, we proposed to upgrade our air bag requirements for passenger cars and light trucks to meet the twin goals mandated by the Transportation Equity Act for the 21st Century: improving protection for occupants of all sizes, belted and unbelted, in moderate to high speed crashes; and minimizing the risks posed by air bags to infants, children, and other occupants, especially in low speed crashes. In response to the public comments on our 1998 proposal and to other new information obtained since issuing the proposal, we are issuing a supplemental proposal that updates and refines the amendments under consideration.

With respect to the goal of improving protection, we are proposing to adopt one of the following alternative crash tests to evaluate the protection of unbelted occupants in moderate to high speed crashes, i.e., those that are potentially fatal. One alternative is an unbelted rigid barrier test (perpendicular and up to + 30 degrees oblique to perpendicular) with a maximum speed to be established in the final rule within the range of 40 to 48 km/h (25 to 30 mph). If we reduce the maximum speed to 40 km/h (25 mph) permanently, we might also increase the maximum speed of the belted rigid barrier test from the current 48 km/h to 56 km/h (30 to 35 mph). Another alternative is an unbelted offset deformable barrier test with a maximum speed to be established in the final rule within the range of 48 to 56 km/h (30 to 35 mph). The vehicle would have to meet the requirements both in tests with the driver side of the vehicle engaged with the barrier and in tests with the passenger side engaged.

With respect to the goal of minimizing the risks of air bags in low speed crashes, we continue to propose performance requirements to ensure that future air bags do not pose unreasonable risk of serious injury to out-of-position occupants. We continue to propose to adopt a number of options for complying with those requirements so that vehicle manufacturers would be free to choose from a variety of effective technological solutions and to develop new ones if they so desire. With this flexibility, they could use technologies that modulate or otherwise control air bag deployment so deploying air bags do not cause serious injuries, technologies that prevent air bag deployment if children or out-of-position occupants are present, or a combination thereof.

DATES: You should submit your comments early enough to ensure that Docket Management receives them not later than December 30, 1999.

ADDRESSES: You may submit your comments in writing to: Docket Management, Room PL-401, 400 Seventh Street, S.W., Washington, D.C., 20590. You may also submit your comments electronically by logging onto the Dockets Management System website at http://dms.dot.gov Click on "Help & Information" or "Help/Info" to obtain instructions for filing the document electronically. Regardless of how you submit your comments, you should mention the docket number of this document.

You may call Docket Management at 202-366-9324 and visit the Docket from 10:00 a.m. to 5:00 p.m., Monday through Friday.

FOR FURTHER INFORMATION CONTACT:

For information about air bags and related rulemakings: Visit the NHTSA web site at http://www.nhtsa.dot.gov and select "Air Bags" under "Popular Information."

For non-legal issues, you may contact Clarke Harper, Chief, Light Duty Vehicle Division, NPS-11. Telephone: (202) 366-2264. Fax: (202) 366-4329. E-mail: Charper@NHTSA.dot.gov.

For legal issues, you may contact Edward Glancy, Office of Chief Counsel, NCC-20. Telephone: (202) 366-2992. Fax: (202) 366-3820.

You may send mail to both of these officials at the National Highway Traffic Safety Administration, 400 Seventh St., S.W., Washington, D.C., 20590.

SUPPLEMENTARY INFORMATION:

Note to readers: As an aid to readers who are outside the engineering community, we have provided at the end of this document a glossary that briefly explains the key technical terms used in this preamble. In the case of the term, "fixed barrier crash test," we have supplemented the explanation with illustrations. That glossary appears in Appendix B. Interested persons may find it helpful to review that glossary before reading the rest of this document.

Table of Contents

  1. Executive Summary.
  2. Background.
    1. Statutory Requirements.
    2. Existing Air Bag Requirements.
    3. September 1998 NPRM.
    4. Public Comments
      1. Tests for Requirements to Improve Occupant Protection for Different Size Occupants, Belted and Unbelted.
        1. Belted Rigid Barrier Test.
        2. Unbelted Rigid Barrier Test.
        3. Up-to-40 km/h (25 mph) Offset Deformable Barrier Test.
      2. Tests for Requirements to Minimize the Risk to Infants, Children and Other Occupants from Injuries and Deaths Caused by Air Bags.
        1. Tests to Minimize Risks to Infants.
        2. Tests to Minimize Risks to Children.
        3. Tests to Minimize Risks to Adults.
      3. Injury Criteria.
    5. Events since September 1998
  3. SNPRM for Advanced Air Bags.
    1. Introduction.
    2. Existing and Proposed Test Requirements.
      1. Tests for Requirements to Improve Occupant Protection for Different Size Occupants, Belted and Unbelted.
        1. September 1998 NPRM.
        2. Comments on September 1998 NPRM.
        3. SNPRM
          1. Requirements for Tests with Unbelted Dummies.
          2. Proposed Array of Crash Test Requirements.
          3. Location and Seating Procedures for 5th Percentile Adult Female Dummy.
      2. Tests for Requirements to Minimize the Risk to Infants, Children and Other Occupants from Injuries and Deaths Caused by Air Bags.
        1. Safety of Infants.
        2. Safety of Young Children.
        3. Safety of Small Teenage and Adult Drivers.
    3. Injury Criteria.
      1. Head Injury Criteria.
      2. Neck Injury Criteria.
      3. Thoracic Injury Criteria.
      4. Lower Extremity Injury Criteria.
      5. Other Criteria.
    4. Lead Time and Proposed Effective Date.
      1. Large Manufacturers.
      2. Small Manufacturers and Multi-stage Manufacturers.
    5. Availability of Original Equipment and Retrofit Manual On-Off Switches.
    6. Warning Labels and Consumer Information.
    7. Miscellaneous Issues.
      1. Selection of Child Restraints.
      2. Due Care Provision.
      3. Selection of Options.
      4. Relationship of Proposed New Injury Criteria to Existing Test Requirements.
      5. Time Parameters for Measuring Injury Criteria During Tests.
      6. Cruise Controls.
      7. Rescue Operations.
      8. Assessing Lower Extremity Injury Potential in Offset Deformable Crash Tests.
      9. Hybrid III Dummy Neck.
    8. Relationship between the NPRM, Comments on the NPRM and this SNPRM.
  4. Costs and Benefits.
  5. Rulemaking Analyses and Notices.
  6. Submission of Comments.
Proposed Regulatory Text.
Appendix A--Response to Petition.
Appendix B--Glossary

I.        Executive Summary.

Since the early 1990's, NHTSA has been taking steps to reduce the risk that air bags will sometimes cause deaths, particularly to unrestrained children and small adults, and to maintain and improve the benefits of air bags. Our initial efforts to reduce the risks focused on a public education campaign to alert the public about the dangers of air bags to children in general and to infants in particular. We urged parents to place their children in the back seat whenever possible and to ensure that they were always properly restrained.

Later, to speed the redesigning and recertifying of air bags that reduce the risks to out-of-position occupants, we established a temporary option allowing vehicle manufacturers to certify their vehicles based on an unbelted sled test. The sled test is simpler, less expensive, and easier to meet than the pre-existing 30 mph unbelted crash test. Limited available data appear to indicate that these redesigned air bags have reduced the risks from air bags for the at-risk populations. However, it is not possible at this time to draw statistically significant conclusions about this.

There is a greater amount of data on the overall benefits of air bags. These data indicate that the redesigned air bags (1) provide essentially the same protection as that provided by earlier air bags. We have considered this information in light of agency tests showing that most of the tested vehicles, although certified to the sled tests, also passed the more stringent 30 mph unbelted crash test.

Manufacturers are developing an assortment of technologies, commonly referred to as advanced air bag technologies, to reduce the risks still further, for children, as well as adults. These technologies include dual-stage inflators which enable air bags to inflate with two different levels of power and which can be linked to various types of sensors including those that sense crash severity, belt use, and seat position (i.e., the location of a vehicle seat on its track). Occupant weight sensors and pattern sensors can be used to prevent an air bag from deploying at all in the presence of children.

These advanced air bag technologies are not just hypothetical possibilities; vehicle manufacturers are beginning to install them in an increasing variety of vehicles. The MY 1999 Hyundai Sonata has a weight sensor designed to prevent the passenger air bag from deploying unless a weight of more than 66 pounds is detected on the passenger seat. Honda introduced a dual stage inflator in its MY 1999 Acura. The MY 2000 Ford Taurus and Honda Accord, which are among the highest selling models in this country, have dual-stage air bags. Some luxury vehicles also have advanced air bag technologies. For example, Mercedes and BMW have dual-stage air bags in some of their MY 2000 cars. The MY 2000 Cadillac Seville has weight and pattern sensors in the passenger seat that work together to turn off the passenger air bag when children are present.

In the Transportation Equity Act for the 21st Century (TEA 21), (2) Congress mandated that we issue a final rule that requires the installation of air bags meeting, by means that include advanced air bag technologies, two goals: first, improving occupant protection for occupants of different sizes, regardless of whether they use their seat belts, and second, minimizing the risk to infants, children and other occupants of deaths and injuries caused by air bags. In accordance with TEA 21, we published a proposal in September 1998 to require the timely introduction of advanced air bags by all vehicle manufacturers and to establish procedures for testing the risk-reducing capabilities of the various types and combinations of advanced air bag technologies. Given the twin goals mandated by TEA 21, the proposal was necessarily both expansive and complex.

To meet the first goal of improving occupant protection, we proposed a variety of tests using belted and unbelted dummies. We also proposed adding a new dummy representing short-statured adult females. Included in these proposals was a proposal to terminate the unbelted sled test option so that vehicles with advanced air bags would be tested in unbelted barrier crashes. The sled test option was valuable as a short-run expedient to make it easier for manufacturers to bring redesigned air bags to market quickly. However, for the long-run purpose of testing air bags to ensure that they are, and that they will continue to be, effective in protecting people in real world crashes, the agency tentatively concluded that air bags should be evaluated in tests simulating those crashes. In particular, the agency proposed to rely on an unbelted 48 km/h (30 mph) rigid barrier crash test that approximates many of the real world crashes severe enough to pose significant risk of serious or fatal injury. Among the tests for belted occupants was a new 40 km/h (25 mph) offset deformable barrier test which was intended to evaluate the ability of crash sensors to sense soft pulse crashes.

With respect to the second goal of minimizing the risks of air bags, the very breadth of the different technological approaches for meeting that goal necessitated we make our proposal even more expansive and complex. We proposed to adopt in the final rule an array of tests to accommodate these different technological approaches and the different choices being made by individual manufacturers about which types of those technologies to adopt. In some cases, we were able to propose generic tests that are suitable for all advanced air bags. In other cases, however, we had to propose tests that are tailored to particular technologies and that would apply to only those air bags incorporating those technologies. This array of tests was intended to provide the manufacturers with technology and design flexibility, while providing the agency with effective means of evaluating the performance of all of the different advanced air bag systems.

The public comments and the agency research and analysis since our 1998 NPRM have enabled us to refine and in some cases simplify the proposed amendments that we are considering. In view of the importance of some of the changes, we have decided to publish this SNPRM to obtain further public comment before making any final decisions and issuing a final rule.

We have reduced the number of proposed dynamic and static tests, especially those relating to the proposed requirements for reducing the risks of air bags. We have reduced, from 14 to nine, (3) the number of proposed dynamic crash tests that would be applicable to all vehicles. We originally proposed that vehicles equipped with static air bag suppression systems (e.g., weight sensors and pattern sensors) be subject to being tested with any child restraint manufactured over a ten-year period. This would have created the possibility of testing with any one of several hundred different models of child restraints. Recognizing that, we solicited comments to aid us in identifying a much more limited number of specific models that would be representative of the array of available child restraints. Based on the public comments, we are now proposing to require that vehicles be able to meet the applicable requirements when tested with any one of a far more limited number of child restraints representing a cross-section of the restraints currently on the market. (4) We have also significantly reduced the number of positions in which test dummies or child restraints could be placed for testing a static suppression system. This was accomplished largely by eliminating positions that were substantially similar to other positions.

We are proposing to expressly provide that manufacturers may use children or small women instead of dummies in static tests to provide a basis for certifying compliance with the proposed tests for static suppression systems. These are simple tests in which the vehicle does not move, and the air bags cannot deploy. We are making this proposal because existing anthropomorphic test dummies were not designed to replicate the weight distribution of sitting humans in a manner that would adequately test all suppression technologies, e.g., pressure/pattern recognition sensors in the vehicle seat. Since the ultimate goal of our provisions concerning suppression systems is to achieve high reliability in detecting the presence of humans, the use of humans for the simple and limited purpose of testing the static suppression systems would make good sense. It is unnecessary to propose the use of infants for certification purposes, since all of the infant restraints should be detectable by any suppression system, regardless of whether they are occupied by a dummy or an infant.

We have eliminated the proposed test for dynamic automatic suppression systems (DASS) and the proposed full scale out-of-position test including pre-crash braking. Public comments and our further testing have led us to conclude that these tests would require enhancements to dummy biofidelity and test procedure development that we could not complete in time for this rulemaking. Further, the commenters did not suggest any workable, effective tests that we could propose as replacements.

Instead, we are taking a different approach that will provide flexibility to manufacturers that may wish in the future to certify advanced air bag systems incorporating a DASS to Standard No. 208. We believe that it is important in crafting our proposals regarding advanced air bags to facilitate efforts by the manufacturers to develop new and possibly better ways of reducing air bag risks. Accordingly, we are proposing to establish very general performance requirements for DASS and a special expedited petitioning and rulemaking process for considering procedures for testing advanced air bags incorporating one of these systems. Target time limits for each phase of such a rulemaking are proposed. Anyone wishing to market such advanced air bags could develop test procedures for demonstrating the compliance of their particular DASS with the performance requirements and submit those test procedures to the agency for its consideration. If the agency deems it appropriate to do so after evaluating the petition, the agency would publish a notice proposing to adopt the manufacturer's test procedure. After considering those comments, the agency would then decide whether the procedure should be added to Standard No. 208. If it decided to do so, and if the procedure were suitable for the DASS of any other vehicles, then the procedure could be used by those manufacturers of those vehicles as well as by the petitioning manufacturer. The agency intends to minimize the number of different test procedures that are adopted for DASS and to ensure ultimately that similar DASS are tested in the same way.

We have also decided to change our proposed injury criteria. We have decided to drop our proposal for a new combined thoracic index (CTI) and instead maintain separate limits for thoracic acceleration and deflection. (5) While CTI may be a better predictor of thoracic injury than chest acceleration and chest deflection independently, there is debate in the biomechanics community about the interpretation of the data. Consequently, we are pursuing further research to resolve the issues.

We are also proposing to change the existing head injury criterion (HIC) for the 50th percentile adult male dummy. (6) HIC is currently required not to exceed 1,000 and is evaluated over a 36 millisecond period. We are proposing to evaluate the HIC over a maximum 15 millisecond time interval with a requirement that it not exceed a maximum of 700. The agency historically has used a 36 millisecond time interval to measure HIC primarily because this method allowed the HIC measurement to indirectly capture risk of neck injury (until recently, a direct indication of neck injury risk was not a part of Standard 208). With the addition of specific neck injury criteria to Standard 208, the agency can switch to a 15 ms measurement interval which better corresponds to the underlying biomechanical research. We are proposing to change the HIC time interval to a maximum of 15 milliseconds for all dummy sizes and to revise the HIC limits by commensurate amounts, based on a scaling from the proposed new limit for the 50th percentile adult male dummy.

We are proposing a neck injury criteria (Nij) limit of 1.0, the calculation of which has been revised since the NPRM. In the NPRM, we requested comments on performance limits of Nij=1 and Nij=1.4. After considering the comments, the available biomechanical data, and testing which indicates that the more conservative or stringent value of 1.0 can be met in current production vehicles, we are proposing a limit of 1.0. The formulae underlying the calculation of Nij for smaller dummies incorporate scaling in recognition of the greater susceptibility of children to injury.

Finally, we are proposing two alternative crash tests for evaluating the effectiveness of an advanced air bag in protecting unbelted occupants in a relatively high speed crash. These tests would be conducted with dummies representing 50th percentile adult males as well as with ones representing 5th percentile adult females. We contemplate adopting one of these tests in a final rule, although we could decide to require elements of both alternatives. We believe that crashing a complete vehicle into a barrier is needed to address the type of situation for which air bags are designed: frontal crashes involving vehicles striking another object with sufficient force that the impact of an occupant with the steering wheel, dashboard, or other interior surface could result in severe injuries or death.

The first alternative is an unbelted rigid barrier test (perpendicular and up to + 30 degrees oblique to perpendicular) with a maximum speed to be established in the final rule within the range of 40 to 48 km/h (25 to 30 mph). This alternative is similar to the test included in our 1998 NPRM. The agency's intent in this rulemaking is to maximize, to the extent consistent with TEA 21, the protection that air bags offer in crashes potentially resulting in fatal injuries. Thus, the agency's preference is to establish such a test requirement at as high a severity as practicable. The 40 km/h (25 mph) lower end of the maximum test speed range is set forth for comment in this notice to ensure that commenters address a crash test recommended by the Alliance of Automobile Manufacturers in late August 1999. If we reduce the maximum speed to 40 km/h (25 mph) permanently, we might increase the maximum speed of the belted rigid barrier test from the current 48 km/h to 56 km/h (30 to 35 mph). The increase could go into effect after the TEA 21 phase-in period.

The second alternative is an unbelted offset deformable barrier test with a maximum speed to be established in the final rule within the range of 48 to 56 km/h (30 to 35 mph). The vehicle would have to meet the requirements both in tests with the driver side of the vehicle engaged with the barrier and in tests with the passenger side engaged. As in the case of the first alternative, if the agency selected this second alternative for the final rule, it would establish the maximum speed at as high a level as practicable, consistent with TEA 21, to maximize the improvement in occupant protection in potentially fatal crashes.

Regardless of which unbelted test or tests we ultimately adopt, we would retain a belted rigid barrier test with a maximum speed of 48 km/h (30 mph) with both 50th percentile adult male and 5th percentile adult female dummies during the TEA 21 phase-in period. (7) Further, we are continuing to propose an up-to-40 km/h (25 mph) offset deformable barrier test requirement, using belted 5th percentile adult female dummies.

We are also continuing to propose to eliminate provisions which allow original equipment (OE) and retrofit on-off switches under specified circumstances. Instead of proposing to phase these provisions out as advanced air bags are phased in, as proposed in the NPRM, we are proposing to allow OE and retrofit on-off switches to be installed under the same conditions that currently apply for all vehicles produced prior to September 1, 2005, the date by which all vehicles must have an advanced air bag system. We believe that by that time consumer confidence in the advanced air bag systems will be sufficiently strong to remove any desire for a manual on-off switch in vehicles produced with an advanced air bag.

NHTSA is proposing a replacement for the permanent sun visor label for vehicles certified as meeting the requirements of this proposed rule. The label would have new graphics and contain statements regarding belt use and seating children in the rear seat. In addition, we are proposing a new temporary label that states that the vehicle meets the new requirements for advanced air bags. This label would replace the existing temporary label and include statements regarding seat belt use and children in rear seats.

II.        Background

A.        Statutory Requirements.

As part of TEA 21, Congress required us to issue an NPRM and final rule meeting two different, equally important goals:

to improve occupant protection for occupants of different sizes, belted and unbelted, under Federal Motor Vehicle Safety Standard No. 208, while minimizing the risk to infants, children, and other occupants from injuries and deaths caused by air bags, by means that include advanced air bags.

(Emphasis added.) (8)

The Act provided that we were to issue the final rule by September 1, 1999. However, if we determined that the final rule could not be completed by that date, the Act provided that the final rule could be issued as late as March 1, 2000. Because of the complexity of the issues and the need to issue this SNPRM, we determined that the final rule could not be completed by September 1, 1999. Under the Act, the final rule must therefore be issued by March 1, 2000.

TEA 21 addressed various other issues, including the effective date for the final rule. A complete discussion of the Act's provisions is included in the 1998 NPRM. See 63 FR 49961.

B.        Existing Air Bag Requirements.

Pursuant to a provision in the Intermodal Surface Transportation Efficiency Act of 1991 (ISTEA), Standard No. 208 requires all passenger cars and light trucks to provide automatic protection by means of air bags. (9)

The automatic protection requirements are performance requirements. The standard does not specify the design of an air bag. Instead, when tested under specified test conditions, vehicles must meet specified limits for injury criteria, including criteria for the head, chest and thighs, measured on 50th percentile adult male test dummies.

Until recently, these criteria limits had to be met for air bag-equipped vehicles in barrier crashes at speeds up to 48 km/h (30 mph), both with the dummies belted and with them unbelted. However, on March 19, 1997, we published a final rule providing manufacturers with the option of certifying the air bag performance of their vehicles with an unbelted dummy in a sled test incorporating a 125 millisecond standardized crash pulse instead of in a vehicle-to-barrier crash test. We made this amendment primarily to expedite manufacturer efforts to reduce the force of air bags as they deploy.

Under the March 1997 final rule, the sled test option was scheduled to terminate on September 1, 2001. We believed there was no need to permanently reduce Standard No. 208's performance requirements, since a variety of longer term alternatives were available to manufacturers to address adverse effects of air bags.

The September 1, 2001 termination date for the sled test option was superseded by a provision in TEA 21. In a paragraph titled "Coordination of Effective Dates," the Act provides that the unbelted sled test option "shall remain in effect unless and until changed by [the final rule for advanced air bags]."

C.        September 1998 NPRM.

Pursuant to TEA 21, on September 18, 1998, we published in the Federal Register (63 FR 49958) a notice of proposed rulemaking (NPRM) to upgrade Standard No. 208, Occupant Crash Protection, to require vehicles to be equipped with advanced air bags that meet new, more rigorous performance requirements. The advanced air bags would be required in some new passenger cars and light trucks beginning September 1, 2002, and in all new cars and light trucks beginning September 1, 2005.

As we explained in that document, air bags have been shown to be highly effective in saving lives. They reduce fatalities in frontal crashes by about 30 percent. However, they also sometimes cause fatalities to infants in rear facing child safety seats and out-of-position occupants.

In the 1998 NPRM, we presented a full discussion of the safety issues related to air bags. We also presented a discussion of our comprehensive plan to address air bag fatalities, which includes requiring advanced air bags as a long-term solution.

We proposed to add a new set of requirements to prevent air bags from causing injuries and to improve the protection that they provide occupants in frontal crashes. There would be several new performance requirements to ensure that the advanced air bags do not pose unreasonable risks to out-of-position occupants.

The NPRM gave alternative options for complying with those requirements so that vehicle manufacturers would be free to choose from a variety of effective technological solutions and to develop new ones if they so desire. With this flexibility, they could use technologies that modulate or otherwise control air bag deployment so deploying air bags do not cause serious injuries or that prevent air bag deployment if children or out-of-position occupants are present.

To ensure that the new air bags are designed to avoid causing injury to a broad array of occupants, we proposed test requirements using dummies representing 12-month-old, 3-year-old and 6-year-old children, and 5th percentile adult females, as well as tests representing 50th percentile adult males. We noted that many of the proposed test procedures were new, and specifically requested comments with respect to their suitability for measuring the performance of the various advanced systems under development.

We also proposed requirements to ensure that the new air bags are designed to cushion and protect an array of belted and unbelted occupants, including teenagers and small women. The standard's current dynamic crash test requirements specify the use of 50th percentile adult male dummies only. We proposed also to specify use of 5th percentile adult female dummies in dynamic crash tests. The weight and size of these dummies are representative of not only small women, but also many teenagers.

In addition to the existing rigid barrier test, representing a relatively "stiff" or "hard" pulse crash in perpendicular tests and a more moderate pulse crash in oblique tests, we proposed to add a deformable barrier crash test, representing a relatively "soft" pulse crash. This proposed new crash test requirement was intended to ensure that air bag systems are designed so that they do not deploy too late. Some current air bags deploy relatively late in certain types of crashes. If an air bag deploys too late, normally seated occupants may move too close to the air bag before it starts to inflate. In such a situation, the air bag is less likely to protect the occupant and may pose a risk to the occupant. We proposed to use 5th percentile adult female dummies in this test.

We also proposed to phase out the unbelted sled test option as we phased in requirements for advanced air bags. We acknowledged that the sled test option has been an expedient and useful temporary measure to ensure that the vehicle manufacturers could quickly redesign all of their air bags and to help ensure that some protection would continue to be provided. Nevertheless, we stated that did not consider sled testing to be an adequate long-term means of assessing the extent of occupant protection that a vehicle and its air bag will afford occupants in the real world.

Finally, we proposed new and/or upgraded injury criteria for each of the proposed new test requirements, and also proposed to upgrade some of the injury criteria for the standard's existing test requirements.

D.        Public Comments.

We received comments from a wide range of interested persons including vehicle manufacturers, air bag manufacturers, insurance companies, public interest groups, academia, and government. Commenters generally supported the goals mandated by TEA 21--improving the benefits of air bags, while minimizing risks from air bags--but expressed widely differing views as to how to accomplish those goals.

In this section of the preamble, we summarize the comments, particularly those relating to the major issues. Because of the large number of public comments, we have included a representative sample of the comments and the commenters who made them.

1.        Tests for Requirements to Improve Occupant Protection for Different Size Occupants, Belted and Unbelted.

a.        Belted Rigid Barrier Test.

A number of vehicle manufacturers opposed adding a belted rigid barrier test using 5th percentile adult female dummies. These commenters argued that this particular test is redundant given the existing belted barrier test using 50th percentile adult male dummies and the other proposed tests using 5th percentile adult female dummies.

The comments of the vehicle manufacturers on this issue were reflective of a more general theme running through their comments, i.e., they believed the NPRM was overly complex and included too many tests.

b.        Unbelted Rigid Barrier Test.

Commenters had sharply different views on our proposal to phase out the unbelted sled test option and reinstate the up-to-48 km/h (30 mph) unbelted rigid barrier test. Many commenters, including all vehicle manufacturers and the Insurance Institute for Highway Safety (IIHS), strongly opposed reinstating the unbelted rigid barrier test. These commenters generally argued that reinstating this test would necessitate a return to "overly aggressive" air bags and that the test is not representative of typical real world crashes. Vehicle manufacturers requested that the sled test option remain available for the long term. On the issue of possible alternative unbelted tests, IIHS suggested that, if we wish to phase out the sled test, we should consider replacing it with a 56 km/h (35 mph) offset deformable barrier test.

On August 31, 1999, however, vehicle manufacturers and their trade associations, Alliance and AIAM, announced to the agency a recently reached consensus recommendation for an unbelted crash test. The industry recommended an unbelted rigid barrier crash test at 40 km/h (25 mph) using both 50th percentile adult male dummies and 5th percentile adult female dummies. The test would be conducted in the perpendicular mode only, i.e., there would be no oblique tests. No supporting data or written analyses were submitted to the agency at that meeting.

Other commenters, including a number of advocacy groups, argued that the up-to-48 km/h (30 mph) unbelted rigid barrier test is representative of a significant portion of real world crashes, and that improvements in vehicle and air bag designs will enable manufacturers to meet the test without safety tradeoffs. Public Citizen argued that while the manufacturers attempt to blame the unbelted barrier test for the deaths and injuries caused by air bags, a closer examination suggests that manufacturers' design selection is the real cause of injuries. It further argued that TEA 21 contemplates that neither belted occupants nor unbelted occupants be favored under Standard 208 and that both deserve safe and effective protection by air bags.

c.        Up-to-40 km/h (25 mph) Offset Deformable Barrier Test.

Commenters' views on the proposed up-to-25-mph belted offset deformable barrier test were mixed, but mostly supportive. Many commenters, including several advocacy groups and a number of vehicle manufacturers, supported the addition of an offset deformable barrier test.

Some vehicle manufacturers requested that the test be conducted only with the driver's side engaged, instead of with either side engaged as proposed in the NPRM. The Association of International Automobile Manufacturers (AIAM) stated that a test with the driver's side engaged would more likely produce "worst case" driver out-of-position locations and possible driver-side intrusion, and that a passenger side offset test would be redundant. Another suggestion made by some vehicle manufacturers was to conduct the test only at 40 km/h (25 mph), rather than at speeds up to 40 km/h (25 mph).

General Motors (GM) stated that it agreed with the addition of the offset deformable barrier test only if the unbelted sled test option remained in effect. GM stated that the offset deformable barrier test augments the sled test by addressing the crash sensing aspects of performance.

DaimlerChrysler argued that the addition of a 40 km/h (25 mph) belted offset deformable barrier test for the 5th percentile female is unnecessary in light of future "depowered" and/or advanced air bags. That commenter stated that injury risks to small occupants sitting near the driver air bag are adequately assessed using the proposed out-of-position, low-risk deployment tests, which it endorses.

Some vehicle manufacturers indicated that air bags might be designed so that they would not deploy in 40 km/h (25 mph) offset crashes.

2.        Tests for Requirements to Minimize the Risk to Infants, Children and Other Occupants from Injuries and Deaths Caused by Air Bags.

a.        Tests to minimize risks to infants.

While commenters generally supported adding tests for infant safety, they raised a number of issues about the proposed tests.

The vehicle manufacturers opposed the proposal to test with any infant seat manufactured during approximately the 10 years prior to the date of vehicle manufacture, citing practicability concerns. A number of vehicle manufacturers also argued that the agency proposed too many test positions. Commenters raised numerous concerns about the specific details of the proposed test procedures.

Some commenters suggested that the agency require suppression in the presence of infants, instead of permitting a low-risk deployment option as well. These commenters cited uncertainties related to injury risk for infants and the lack of infant biomechanical data. They further questioned if there is any benefit from air bag deployments for infants.

A number of commenters also raised concerns about whether suppression devices will be ready in time to meet the requirements for advanced air bags, and how reliable they will be.

b.        Tests to minimize risks to children.

Commenters' views on the proposed tests for child safety were similar to those for infant safety. While supportive of adding tests in this area, vehicle manufacturers raised concerns about the number of child restraints, number of tests, and, in some cases, availability of reliable suppression devices.

A number of commenters raised concerns about whether current child dummies are sufficiently human-like to be appropriate test devices for some of the advanced technologies under development. By way of example, concern was expressed that suppression devices that work by sensing the distributed weight pattern of a child on a seat may not recognize the pattern of a test dummy.

Commenters raised numerous technical issues concerning the proposed options for automatic suppression features that suppress the air bag when an occupant is out-of-position (S27 of the regulatory text proposed in the NPRM). Some commenters argued that the proposal to test automatic suppression features using a moving headform is not appropriate for some of the devices under development, such as sensors designed to track the full body of the occupant and not just the head. Others expressed difficulties related to defining the size, shape, and orientation of the suppression plane, as well as the maximum response time of the system.

Commenters also raised numerous technical issues concerning the dynamic out-of-position test (S29 of the regulatory text proposed in the NPRM). Some commenters stated that the dummy trajectories resulting in this test are unrealistic, and that the proposed vehicle crash test is neither repeatable nor reproducible. Others stated that the dummies do not move close enough to the air bag prior to deployment to represent a worst case out-of-position situation.

c.        Tests to minimize risks to adults.

Commenters generally supported adding a low-risk deployment test using a 5th percentile adult female dummy at the driver seating position, although they raised a number of issues about the proposed test procedure. GM recommended that the driver low risk deployment test be made into a component test, outside of the vehicle.

Commenters also raised the same concerns about the proposed options for automatic suppression features that suppress the air bag when an occupant is out-of-position (S27) and for the dynamic out-of-position test (S29) as they did in the context of tests to minimize risks to children.

GM recommended that the agency also propose a low-risk deployment test using a 5th percentile adult female dummy at the passenger position. That company noted that if manufacturers selected the suppression (presence) option for child safety, there would be no out-of-position test limiting aggressivity for adult passengers.

3.        Injury Criteria.

Commenters raised numerous highly technical issues concerning several of proposed injury criteria and performance limits. Some commenters questioned the biomechanical basis for certain of the proposed new injury criteria. The AAMA suggested essentially a completely revised set of injury criteria.

E.        Events since September 1998

A number of events relevant to this rulemaking have occurred since publication of the NPRM in September 1998. First, the development of advanced air bags by suppliers and vehicle manufacturers has continued.

Acura introduced dual stage passenger side air bags in its MY 1999 Acura RL. According to Acura's press release, "(t)he dual stage air bags were designed to reduce the inflation speed to help protect children or small-framed adults. In a low speed collision, the dual-stage inflator system is triggered in sequence resulting in slower air bag deployment with less initial force. In higher speed collisions, both inflators operate simultaneously for full immediate inflation. The air bag system logic also controls the operation of the seat belt pretensioners. A new feature of the system detects whether the passenger's seat belt is fastened. If the seat belt is not fastened, the air bag deploys at full force at a lower collision speed to help offer more protection to the unbelted occupant."

Ford publicly announced in January 1999 that it will introduce advanced technology enabling its cars and trucks to analyze crash conditions and to use the results of the analyses in activating safety devices to better protect a range of occupants in a variety of frontal crash situations. Ford stated that its Advanced Restraints System features nearly a dozen technologically advanced components that work together to give front-seat occupants significantly enhanced protection during frontal crashes, taking into account their seating position, safety belt use and crash severity. That company indicated that elements of the system, which features technologies such as crash severity sensors, a driver-seat position sensor, a passenger weight sensor, safety belt usage sensors, dual-stage inflating air bags, safety belt pretensioners and energy management retractors, will debut in vehicles beginning in the 1999 calendar year. Ford stated that the company will introduce these new technologies on new and significantly freshened models until all its passenger cars, trucks and sport utility vehicles have the complete Advanced Restraints System.

GM publicly announced in February 1999 that it will introduce technology in MY 2000 that is designed to detect the presence of a small child in the front passenger seat and suppress the deployment of the passenger frontal air bag in the event of a frontal crash. GM stated that weight-based sensors, coupled with pattern recognition technology, will distinguish between a child and a small adult female whose weight may be similar to a large child restrained in a child safety seat. If the front passenger seat is occupied by a small child, whether in a child safety seat or not, GM said that the air bag will not deploy. GM stated that it will introduce this technology on the Cadillac Seville in the 2000 calendar year, and that it has a roll-out plan to extend this technology throughout its product line.

We have received more detailed confidential information from GM and Ford concerning their plans, as well as confidential information from other auto manufacturers concerning their latest plans to introduce various advanced technologies. We have also received confidential information from suppliers.

Second, in April 1999, we held a public technical workshop concerning biomechanical injury criteria. The purpose of the workshop was to provide an additional opportunity for a continuing dialog with the biomechanics community and the public to assure that we considered appropriate injury criteria.

Third, we have analyzed the public comments and also conducted additional testing. We conducted additional tests of current vehicles with redesigned air bags to determine how they perform in 48 km/h (30 mph) rigid barrier crash tests. We selected vehicles that varied by class, stiffness, and manufacturer. We also used both 5th percentile adult female dummies and 50th percentile adult male dummies, belted and unbelted. We also conducted tests of several current vehicles with redesigned air bags to determine how they perform in 40 km/h (25 mph) rigid barrier crash tests, 48 km/h (30 mph) 30 degree right/left angular barrier tests (belted/unbelted), 56 km/h (35 mph) left/right side offset fixed deformable barrier crash tests, low speed 24 to 40 km/h (15 to 25 mph) offset deformable crash tests and static out-of-position tests. We also conducted sled tests at different crash severities with 95th percentile adult male dummies and MY 1999 and MY 1997 replacement air bags.

Fourth, we have continued to analyze available data to see how redesigned air bags are performing in the real world. We analyzed 1996 to 1998 Fatality Analysis Reporting System (FARS) data and found essentially the same number of fatalities in frontal impacts for MY 1996 vehicles in 1996 FARS (730), as in MY 1997 vehicles in 1997 FARS (776), as in MY 1998 vehicles in 1998 FARS (732). The fatality rates per million registered vehicles indicate that MY 1996 (56 per million registered vehicles) had essentially the same fatality rates as MY 1997 vehicles (55), while MY 1998 vehicles had a lower fatality rate (50). After controlling for safety belt use rates, that is, estimating the number of fatalities in each year if all three years had the same 1998 usage rate, the fatality rates per million registered vehicles were the same for MY 1996 and MY 1997 (53), while MY 1998 had a lower fatality rate (50). Since an estimated 87 percent of MY 1998 vehicles have redesigned air bags, this suggests that there is essentially the same or slightly better protection provided by the redesigned air bags compared to pre-MY 1998 air bags. In assessing the significance of this information, we will consider the agency tests in which most of the tested vehicles, although certified to the sled tests, met or exceeded the historical performance requirements of the 48 km/h (30 mph) rigid barrier crash test.

Another analysis compared the percent of fatalities in frontal impacts to all impacts for MY 1996 vehicles in calendar year 1996 (38.9%), to MY 1997 vehicles in calendar year 1997 (41.3%), and to MY 1998 vehicles in the first 6-months of calendar year 1998 (39.6%). As noted above, most of the MY 1998 vehicles have redesigned air bags. No statistically significant difference was found between the three sets of data. Again, this implies that the overall protection provided by the redesigned air bags is essentially the same as that provided by pre-MY 1998 air bags.

Fifth, on August 31, 1999, and again on September 14, 1999, the vehicle manufacturers and their trade associations met with the agency and presented a consensus recommendation for an unbelted crash test. The industry recommended an unbelted rigid barrier crash test at 40 km/h (25 mph) using both 50th percentile adult male dummies and 5th percentile adult female dummies. A letter regarding this recommendation was received from the Alliance (dated September 2, 1999). (10)

In a letter dated September 16, 1999, an assortment of commenters, including vehicle manufacturers, vehicle insurers, the American Automobile Association, the National Automobile Dealers Association, the American International Automobile Dealers Association, the American Trauma Society, the National Safety Council, IIHS, and the National Association of Governors' Highway Safety Representatives, opposed a return to the 30 mph unbelted rigid barrier test. This letter argued that a return to this test would require an overall increase in air bag maximum energy levels with a concomitant increase in risk. No supporting data or analysis accompanied the letter. The letter also urged that NHTSA focus this rulemaking on reducing the risk of air bags to children and others, especially in low speed crashes, as compared to the agency's attempting to increase air bag-related benefits for unbelted occupants in higher speed crashes.

In a letter dated September 29, 1999, Public Citizen, the Center for Auto Safety, and Parents for Safer Air Bags stated that they were "concerned by news reports that a consortium of vehicle manufacturers and insurers is pressing the agency not to reinstate the 30 mph barrier crash test for unbelted occupants." These organizations argued that the industry's position is based on the erroneous premise that protection of unbelted occupants in high-speed collisions causes the bags to be hazardous to small occupants in low-speed collisions. (11) They also argued that abandonment of the unbelted 30 mph unbelted test would obviate the very purpose of the present rulemaking, the development and introduction of advanced air bags, and result in the use of generic "lowest common denominator" systems that can be readily be fitted in any vehicle but which seriously compromise safety. The letter stated that it should not be forgotten that air bags were originally conceived to protect unbelted occupants in horrific frontal collisions, and that this remains their principal efficacy to this day.

III.        SNPRM for Advanced Air Bags.

A.        Introduction.

Our primary goals in this rulemaking continue to be those set for us by TEA 21, i.e., to improve occupant protection for occupants of different sizes, belted and unbelted, while minimizing the risk to infants, children, and other occupants from injuries and deaths caused by air bags. Further, we are seeking to ensure that the needed improvements in occupant protection are made in accordance with the statutory implementation schedule. After carefully reviewing the comments on the NPRM and other available information, we have developed an SNPRM to accomplish these goals.

In developing this SNPRM, we focused on picking the most appropriate tests so that we could reduce the number of originally proposed tests without significantly affecting the benefits of the NPRM. We were persuaded by the commenters that reducing the amount of testing was important, given resource limitations, and the costs to manufacturers associated with certifying vehicles to such a large number of new test requirements. At the same time, we wanted to be sure that the SNPRM includes sufficient tests to ensure that air bags are redesigned to meet the goals mandated by TEA 21.

Given the continued debate over what requirements should be relied upon to ensure protection to unbelted occupants, we also wanted to be sure that we have considered and received the benefit of public comments on the various alternative approaches reflecting the views and information now available to us.

The most significant differences between the NPRM and the SNPRM can be summarized as follows:

The existing tests that would be retained as well as those proposed in this SNPRM are identified in Figures 1a, 1b and 2, below. Figures 1a and 1b show the two alternative sets of test requirements to improve occupant protection for different size occupants, belted and unbelted, in moderate to high speed crashes. Figure 2 shows test requirements to minimize the risk to infants, children, and other occupants from injuries and deaths caused by air bags, especially in low speed crashes.

Figure 1a

Figure 1b

Figure 2

A discussion of the specific proposed test requirements follows. We will first discuss requirements to improve protection for different size occupants, belted and unbelted, and will then discuss requirements to minimize risks from air bags. We also discuss in detail the major differences from the NPRM.

B.        Existing and Proposed Test Requirements.

1.        Tests for Requirements to Improve Occupant Protection for Different Size Occupants, Belted and Unbelted.

a.        September 1998 NPRM.

In the NPRM, we proposed test requirements to improve occupant protection for different size occupants, belted and unbelted. The proposed requirements included rigid barrier tests and offset deformable barrier tests.

Under the proposed rigid barrier test requirements in the NPRM, vehicles would have been required to meet injury criteria performance limits, including ones for the head, neck, chest, and femurs, measured on 50th percentile adult male and 5th percentile adult female test dummies during rigid barrier crash tests at any speed up to 48 km/h (30 mph) and over the range of vehicle-to-crash-barrier angles from -30 degrees to +30 degrees. Tests with 50th percentile adult male dummies would be conducted with the vehicle seat in the mid-track position; tests with 5th percentile adult female dummies would be conducted with the vehicle seats in the full forward position. (13) Vehicles were to meet the injury criteria with belted and unbelted dummies. The purpose of the rigid barrier tests was to help ensure that vehicles protect different size occupants, belted and unbelted, from risk of serious or fatal injury in moderate to high speed crashes.

Under the proposed offset deformable barrier test requirements, vehicles would have been required to meet injury criteria performance limits during an up-to-40 km/h (25 mph) frontal offset deformable barrier test, using belted 5th percentile adult female dummies. The frontal offset test would have been conducted with either the driver side of the vehicle or the passenger side of the vehicle engaged with the barrier. The purpose of this test was to help ensure that vehicle manufacturers design their crash sensing and software systems to adequately address soft and long duration crash pulses.

Our NPRM would have required as many as a total of 14 crash tests to improve occupant protection. This number is based on counting each rigid barrier test specifying use of a particular dummy as three tests, reflecting the assumption that, for typical vehicle and air bag designs, there would be three worst case conditions: 48 km/h (30 mph) at -30 degrees, 48 km/h (30 mph) at 0 degrees, and 48 km/h (30 mph) at +30 degrees. (14)

Our proposed requirements for improving occupant protection in potentially fatal crashes differed from the existing Standard No. 208 in several important respects.

First, vehicles would for the first time be required to be certified to crash test requirements using 5th percentile adult female dummies, which would be seated in the full forward seat track position. Historically, the standard has only specified the use of 50th percentile adult male dummies seated further back.

Second, vehicles would be required for the first time to meet neck injury criteria performance limits in a crash test. Neck injuries are a particular concern for persons sitting close to the air bag.

Third, vehicles would for the first time be required to comply with injury criteria limits in a 40 km/h (25 mph) frontal offset deformable barrier test with belted 5th percentile adult female dummies. The only frontal crash tests previously specified by the standard were rigid barrier tests.

Fourth, we proposed to phase out the unbelted sled test option and return to the up-to-48 km/h (30 mph) unbelted rigid barrier test requirement. (15) However, it would be more than simply returning to the previous test requirement, since the unbelted rigid barrier test would now be conducted with 5th percentile adult female dummies as well as 50th percentile adult male dummies. In addition, we proposed added injury criteria for the chest and neck.

We proposed to phase out the sled test option as we phased in the requirements for advanced air bags. We stated that while we believe the sled test option has been an expedient and useful temporary measure to ensure that the vehicle manufacturers could quickly redesign all of their air bags and to help ensure that some protection would continue to be provided by air bags, we did not consider sled testing to be an adequate long-term means of assessing the extent of occupant protection that a vehicle and its air bag will afford occupants in real world crashes.

We noted that the sled test, first, does not address vehicle factors that can significantly affect the level of protection provided in the real world and, second, is not representative of a significant number of potentially fatal real world crashes. Each of these limitations is significant. The first means that sled test results may have limited relationship to real world performance in many types and levels of severity of crash. The second means that sled test results may not be a good measure of air bag performance in the kinds of crashes in which air bags are supposed to save lives. While we proposed to return to the up-to-48 km/h (30 mph) unbelted rigid barrier test requirement, we requested comments on possible alternative unbelted crash test requirements.

b. Comments on 1998 NPRM.

Our proposal to reinstate the up-to-48 km/h (30 mph) unbelted rigid barrier test requirement was by far the most extensively debated issue of this rulemaking. As noted earlier, commenters had sharply different views on this aspect of the NPRM. In their initial comments, motor vehicle manufacturers and their trade associations strongly opposed returning to the up-to-48 km/h (30 mph) unbelted rigid barrier test and urged that the sled test option remain in effect permanently. They argued that reinstating the up-to-48 km/h (30 mph) unbelted rigid barrier test would prevent continued use of "depowered" air bags and require a return to "overly aggressive" air bags and that the test is not representative of typical real world crashes. They argued that the sled test includes a crash pulse that is more representative of typical real world crashes.

On August 31, 1999, however, vehicle manufacturers and their trade associations presented to the agency a consensus recommendation for an unbelted crash test. The industry recommended an unbelted rigid barrier crash test at 40 km/h (25 mph) using both 50th percentile adult male dummies and 5th percentile adult female dummies. The test would be conducted in the perpendicular mode only, i.e., there would be no unbelted oblique tests. Industry representatives argued that oblique tests are not needed to ensure wide air bags as vehicle manufacturers will provide them in light of other considerations, e.g., general safety considerations, the 48 km/h (30 mph) belted rigid barrier crash testing, and IIHS and European high speed belted offset deformable barrier testing.

In its comments on the NPRM, IIHS also opposed returning to the up-to-48 km/h (30 mph) unbelted rigid barrier test, for reasons similar to those cited by the vehicle manufacturers. However, that organization suggested that if we wish to phase out the sled test, we should consider replacing it with the 56 km/h (35 mph) European offset crash into a deformable barrier, using unbelted dummies, instead of the rigid barrier test. IIHS stated that this configuration would address not only protection in asymmetric crashes, but also some issues of intrusion that are related to restraint system performance, e.g., steering column movement. IIHS also stated that adoption of this test would be in the direction of harmonizing European and U.S. test procedures, the only difference being using unbelted versus belted dummies.

On September 14, 1999, however, IIHS advised us that it now believes that an unbelted 56 km/h (35 mph) offset deformable barrier crash test would be inappropriate. That organization is concerned that including this test in Standard No. 208 might lead to an increase in unintended high-energy air bag deployments, posing risks to out-of-position occupants, because of uncertainties in the sensing and algorithm capabilities in making proper deployment decisions. This potential problem is related to the nature of this crash test. During the initial phase of the test, i.e., during the crushing of the deformable barrier face, vehicles experience a long duration, low magnitude acceleration. The crash pulse in this phase of the test resembles that of a low speed crash. After the vehicle crushes the barrier face and reaches the underlying rigid portion, the remaining phase of the test is similar to a rigid barrier test. IIHS is concerned that because the initial phase of the test results in a crash pulse similar to that experienced in a low speed crash, air bag systems might not be able to distinguish between the offset test and a low speed crash during the time the decision whether to deploy the air bag must be made. If this were the case, an air bag system that was designed to meet an unbelted 56 km/h (35 mph) offset deformable barrier crash test by means of a high-energy air bag deployment might inappropriately provide the same kind of deployment in a low speed crash, thereby posing unnecessary risks to out-of-position occupants.

The Automotive Occupant Restraints Council (AORC), representing manufacturers of air bags and seat belts, stated that while it believes the current sled test option serves a useful purpose, a sled test cannot provide a complete assessment of the crash protection provided by a vehicle/restraint system. That organization stated it believes that to fully assess crash protection for belted and unbelted occupants, barrier crash tests of complete vehicles should be included in the test requirements of Standard No. 208. AORC noted that complete vehicle barrier tests permit the evaluation of the vehicle's structure and its contribution to occupant protection. AORC recommended that additional analysis be conducted concerning what barrier and test conditions should be included in Standard No. 208.

A number of commenters, including several public interest groups, argued that the up-to-48 km/h (30 mph) unbelted rigid barrier test is representative of a significant portion of real world crashes, and that improvements in vehicle and air bag designs will enable manufacturers to meet the test without safety tradeoffs.

As to the proposed belted tests, some vehicle manufacturers argued in their comments on the NPRM that a belted rigid barrier test using 5th percentile adult female dummies would be redundant. They argued that the combination of other tests using 5th percentile adult female dummies plus the existing rigid barrier test using belted 50th percentile adult male dummies would address the same area of safety.

Commenters' views on the proposed up-to-40 km/h (25 mph) belted offset deformable barrier test were mixed, but mostly supportive. Many commenters, including several safety advocacy groups and a number of vehicle manufacturers, supported the addition of an offset deformable barrier test.

As noted earlier, some vehicle manufacturers requested that the test be conducted only with the driver's side engaged, instead of with either side engaged as proposed in the NPRM. The Association of International Automobile Manufacturers (AIAM) stated that a test with the driver's side engaged would more likely produce worst case driver out-of-position locations and possible driver-side intrusion, and that a passenger side offset test would be redundant. Another suggestion made by some vehicle manufacturers was to conduct the test only at 40 km/h (25 mph), rather than at speeds up to 40 km/h (25 mph).

General Motors (GM) stated that it agreed with the addition of the offset deformable barrier test only if the unbelted sled test option remained in effect. GM stated that the offset deformable barrier test augments the sled test by addressing the crash sensing aspects of performance.

DaimlerChrysler argued that the addition of a 40 km/h (25 mph) belted offset deformable barrier test for the 5th percentile adult female is unnecessary in light of future "depowered" and/or advanced air bags. That commenter stated that injury risks to small occupants sitting near the driver air bag are adequately assessed using the proposed out-of-position, low-risk deployment tests, which it endorses.

c.        SNPRM.

We believe that the comments on the proposed test requirements to improve occupant protection for different size occupants, belted and unbelted, raise two primary questions:

(1) What type and severity level of an unbelted crash test should be included in Standard No. 208?

(2) Are some of the tests proposed in the NPRM redundant, given the other proposed tests?

In the sections which follow, we will address what unbelted test requirements are needed to address the protection of unbelted teenagers and adults, and what overall set of requirements is needed to improve protection for different size occupants, belted and unbelted.

i.        Requirements for Tests with Unbelted Dummies

As we address the issue of what unbelted requirements should be included in Standard No. 208 to address the protection of unbelted teenagers and adults, we believe the ultimate question for regulators, industry and the public is how the required safety features work in the real world. We will consider that question as we separately address two issues: (1) sled testing versus crash testing, and (2) alternative unbelted crash tests (e.g., rigid barrier crash tests, offset deformable tests, etc.) at various severity levels.

Crash testing vs. sled testing.

In a full-scale crash test, instrumented test dummies are placed in a production vehicle, and the vehicle is actually crashed. Measurements from the test dummies are used to determine the forces, and injury potential, human beings would have experienced in the crash.

Many different types of crash tests can be conducted, and the various types of crash tests can be conducted at different levels of severity. Commonly conducted crash tests include: (1) rigid barrier tests, in which a vehicle is crashed head-on (perpendicular) or at an angle into a rigid barrier, (2) offset deformable barrier tests, in which a vehicle is crashed into a barrier with a deformable face, with only a portion of the front of the vehicle (e.g., 40 percent) engaging the barrier, and (3) moving deformable barrier tests, in which a moving deformable barrier designed to be representative of particular vehicles is crashed into the test vehicle. Vehicle-to-vehicle crash tests, in which one vehicle is crashed into another vehicle, are sometimes used in research or product development.

In a sled test, no crash takes place. The vehicle is essentially undamaged. The vehicle is placed on a sled-on-rails, and instrumented test dummies are placed in the vehicle. The sled is accelerated very rapidly backwards (relative to the direction that the occupants would be facing), so that the occupant compartment experiences the same motion as might be experienced in a crash. The air bags are manually deployed at a pre-selected time during the sled test. Measurements from the test dummies are used to determine the forces, and injury potential, human beings would have experienced during the test.

In the NPRM, we explained that the agency has long specified full scale vehicle crash tests using instrumented dummies, in a variety of our standards, because it is only through such tests that the protection provided by the vehicle occupant protection system can be fully measured.

In the NPRM, we cited several significant limitations of the current sled test, some of which are inherent to any sled test. We explained:

Unlike a full scale vehicle crash test, a sled test does not, and cannot, measure the actual protection an occupant will receive in a crash. The current sled test measures limited performance attributes of the air bag, but cannot measure the performance provided by the vehicle structure in combination with the air bags or even the full air bag system by itself.

Among other shortcomings, the sled test does not evaluate the actual timing of air bag deployment. Deployment timing is a critical component of the safety afforded by an air bag. If the air bag deploys too late, the occupant may already have struck the interior of the vehicle before deployment begins.

Air bag timing is affected by parts of the air bag system which are not tested during a sled test, i.e., the crash sensors and computer crash algorithm. A barrier crash test evaluates the ability of sensors to detect a crash and the ability of an algorithm to predict, on the basis of initial sensing of the rate of increase in force levels, whether crash forces will reach levels high enough to warrant deployment. However, the sled test does not evaluate these critical factors. The ability of an algorithm to correctly, and quickly, predict serious crashes is critical. The signal for an air bag to deploy must come very early in a crash, when the crash forces are just beginning to be sensed by the air bag system. A delay in an air bag's deployment could mean that the air bag deploys too late to provide any protection. In a sled test, the air bag is artificially deployed at a predetermined time. The time of deployment in a sled test is artificial and may differ significantly from the time when the air bag would deploy during an actual crash involving the same vehicle.

Second, the current generic sled pulse does not replicate the actual crash pulse of a particular vehicle model, i.e., the specific manner in which the front of the vehicle deforms during a crash, thereby absorbing energy. The actual crash pulse of a vehicle is a critical factor in occupant protection. A crash pulse affects the timing of air bag deployment and the ability of an air bag to cushion and protect an occupant. However, the current sled test does not use the crash pulse of the vehicle being tested. In many cases, the crash pulse used in the sled test is not even one approximately representative of the test vehicle. The sled test uses the crash pulse of a large passenger car for all vehicles, regardless of their type or size. This crash pulse is appropriate for large passenger cars, but not for light trucks and smaller cars since they typically have much "stiffer" crash pulses than that of the sled test. In the real world, deceleration of light trucks and smaller cars, and their occupants, occurs more quickly than is simulated by the sled test. Thus, the sled test results may overstate the level of occupant protection that would be provided by a vehicle and its air bag system in the real world. An air bag that can open in a timely fashion and provide adequate cushioning in a soft pulse crash may not be able to do so in a stiffer pulse crash. This is because an occupant of a crashing vehicle moves forward, relative to the vehicle, more quickly in stiffer pulse crash than in a softer pulse crash.

Third, a sled test does not measure the potential for harm from vehicle components that are pushed back into the occupant compartment during a crash. Examples of components that may intrude into the occupant compartment include the steering wheel, an A-pillar and the toe-board. Since a sled test does not involve any kind of crash or deformation of the vehicle, it implicitly assumes that such intrusion does not occur in crashes. Thus, the sled test may indicate that a vehicle provides good protection when, as a result of steering wheel or other intrusion ...., the vehicle will actually provide poor protection in a real world crash.

Fourth, the sled test does not measure how a vehicle performs in angled crashes. It only tests vehicles in a perpendicular crash. In the real world, frontal crashes occur at varying angles, resulting in occupants moving toward the steering wheel and instrument panel in a variety of trajectories. The specification of angled tests in conjunction with the barrier test requirement ensures that a vehicle is tested under these real world conditions. 63 FR 49971.

Commenters supporting retention of the sled test did not dispute the inherent limitations of sled tests as compared to crash tests.

AAMA argued that the single best argument for retaining the existing sled test is that "it's working;" AAMA contended that "depowered" air bags in vehicles certified according to the sled test are saving the lives of occupants of all sizes, while reducing the harm to children and other out-of-position occupants.

It is not clear, however, that the sled test is responsible for any of the benefits of redesigned air bags other than to the extent it made it easier for vehicle manufacturers to redesign and certify their existing air bags more quickly.

As noted earlier, limited available data appear to indicate that redesigned air bags have reduced the risks from air bags for the at-risk populations. However, it is not possible at this time to draw statistically significant conclusions about this. There is a greater amount of data on the overall benefits of air bags. These data indicate that there is essentially the same or slightly better protection provided by the redesigned air bags compared to earlier air bags.

Regardless of how well vehicles with redesigned air bags are currently performing, however, the sled test itself cannot guarantee that future air bags would perform nearly so well. These vehicles and their air bags were initially designed to the unbelted barrier test, and their current air bags represent quick, partial redesigns of those air bags. Thus, their performance is still highly reflective of the unbelted test.

While the sled test has made it easier for manufacturers to redesign and certify their vehicles more quickly, manufacturers could and did depower air bags under Standard No. 208's unbelted barrier test. As discussed below, available data suggest that most vehicles, while certified to the sled test, continue to meet the unbelted barrier test requirements (including the new neck injury criteria) with the 50th percentile adult male dummies.

Our goal in this rulemaking is to determine what requirements to protect unbelted and other occupants should apply to vehicles in the future. AAMA's argument that the sled test is working does not take into account all of the kinds of less protective vehicles and air bags that would be permitted by the sled test, given its mildness, and which might be produced if the sled test were allowed to remain in effect on a long-term basis.

The sled test is unable to offer any assurance that current vehicles and air bags are representative of what manufacturers would offer in the long run if the sled test were available as a permanent option. Nothing in the standard would inhibit manufacturers from making their air bags significantly smaller in both depth and width, and thus less protective in high speed crashes. In particular, narrower air bags could provide less protection in crashes involving oblique angles. The sled test also might permit "face bags" which do not provide chest protection or restraint for portions of the lower torso. In addition, the absence of an unbelted full-vehicle test at an appropriate severity level would permit vehicles to be designed with stiffer, less energy-absorbing front ends, e.g., to provide more interior passenger or cargo-carrying space at the expense of frontal "crush" space.

Moreover, unless balanced by an effective unbelted crash test requirement, the proposed new requirements to minimize air bag risks to out-of-position occupants have the potential to create an incentive for manufacturers to make their current air bags smaller and less protective. An inexpensive and relatively easy way to reduce risks from the air bag to out-of-position occupants is to further depower air bags and make them smaller. However, if air bags are depowered too much or made too small, they will not provide meaningful protection in high speed crashes.

Our basic obligation is to issue Federal motor vehicle safety standards that establish a minimum level of performance that protects the public against unreasonable risk of crashes occurring because of the design, construction, or performance of a motor vehicle, and against unreasonable risk of death or injury in a crash. In this particular rulemaking, we are facing an array of safety problems, and TEA 21 as well as our pre-existing statutory authority, require that we address each of them.

The most reliable way to determine how vehicles will perform in real world crashes is to crash them. That is why we believe that a crash test is needed. Sled tests are useful research tools, but they do not provide as full or accurate a measure of the occupant protection that a vehicle will provide in the real world.

Given the importance of unbelted protection, we believe it is necessary to provide the public with assurance that the minimum level of performance for each vehicle will be required to be meaningful, based on careful scientific and engineering analysis. While we have carefully considered all of the comments concerning the sled test, we continue to believe that sled testing is an inadequate long-term means for ensuring that current levels of unbelted occupant protection are improved. This is based on the above-noted inherent limitations of sled tests, as compared to crash tests, in evaluating occupant protection. Whether one looks at IIHS with its offset crash test program, Europe with its offset NCAP program, or our experience with our NCAP, Standard No. 208 and Standard No. 214, it is widely acknowledged that crash tests, set at appropriate severity levels, provide the best means of evaluating the protection that occupants will receive in real world crashes.

For this SNPRM, we urge commenters to focus on what specific unbelted complete vehicle crash tests are the most appropriate.

Alternative unbelted crash tests

As we noted above, many different types of crash tests can be conducted, and the various types of crash tests can be conducted at different levels of severity and orientation. Commonly conducted crash tests include: (1) fixed rigid barrier tests, (2) fixed offset deformable barrier tests and (3) moving deformable barrier tests.

If government or anyone else wants to determine whether a vehicle provides an appropriate degree of occupant protection in a potentially fatal or serious injury producing crash, the crash test must have the severity representative of those crashes. The fact that a test might indicate that an occupant would not be injured or killed in a relatively mild crash says nothing about whether the occupant would likely be killed in a more serious crash. That is why it is important to distinguish between the universe of all typical real world crashes and those typical real world crashes serious enough to pose a significant risk of serious or fatal injury. While one could argue that the most "typical" crash is probably a fender bender resulting in little or no personal injury, basing Standard No. 208 on such a test would not result in any savings in lives or reductions in serious injuries. Of course, there are many issues to consider in selecting a specific crash test, but we must focus on seeking to represent the kind of typical crashes that are potentially fatal, rather than typical crashes as a whole.

When we issued the NPRM, we released a paper titled "Review of Potential Test Procedures for FMVSS No. 208." The paper provided a detailed technical analysis of the various alternative crash tests. To accompany this SNPRM, we are releasing an updated version of that paper, which has been revised in light of comments and other new information. The paper shows that, among the currently available alternative crash tests, the rigid barrier test (perpendicular and up to + 30 degrees oblique to perpendicular) represents the greatest number of real world crashes involving serious to fatal injuries. The only alternative crash test that would represent a greater number of such crashes would be one involving a moving deformable barrier, which is still undergoing research.

In the NPRM, we noted that while the perpendicular rigid barrier test results in crash pulses of short duration, e.g., the kind of pulse that a vehicle experiences when it fully engages another similar-sized or larger vehicle directly head-on or strikes a bridge abutment, the oblique rigid barrier tests result in crash pulses of longer duration, i.e., a "softer" crash pulse, which may occur when vehicles strike each other at various angles.

We also noted that vehicles and air bags designed to comply with the unbelted rigid barrier test have been effective in saving lives. At the time of the NPRM, we estimated that air bags had saved the lives of about 3,148 drivers and passengers. Of these, 2,267 were unbelted. The rest, 881, were belted. If these levels of effectiveness are maintained (i.e., 21 percent in frontal crashes for restrained occupants and 34 percent in frontal crashes for unrestrained occupants), air bags will save more than 3,000 lives each year in passenger cars and light trucks when all light vehicles on the road are equipped with dual air bags.

Commenters opposing the 48 km/h (30 mph) unbelted barrier test raised two primary issues. First, they argued that the test is not representative of typical crashes. Second, they argued that returning to this test would prevent continued use of "depowered" air bags and would require a return to "overly aggressive" air bags.

We note that, in arguing that the 48 km/h (30 mph) unbelted barrier test is not representative of typical crashes, the commenters did not define what they meant by "typical crashes." Given that the purpose of Standard No. 208 is primarily to reduce serious-to-fatal injuries, we believe that question is whether that test is representative of the crashes that produce those injuries. More than 18,000 drivers and right front passengers are killed each year in frontal impacts, and more than 290,000 drivers and right front passengers experience moderate to critical non-fatal injuries. These numbers would be significantly higher without effective air bags. In order to promulgate safety standards that protect the public against unreasonable risk of death or injury in a crash, and to fulfill our specific duty under TEA 21 to improve occupant protection for occupants of different sizes, belted and unbelted, it is necessary for Standard No. 208 to address these crashes. In addition, by requiring vehicles to provide protection over a range of crash severities, e.g., in tests at speeds "up to" a given velocity, we also address protection for lower severity crashes. The upper level severity must, however, be sufficient to ensure that manufacturers provide life-saving occupant protection in higher speed crashes.

The following figures, derived from National Automotive Sampling System (NASS) data for years 1993-1997, show the cumulative distribution of injuries and fatalities in frontal crashes by delta V, (16) for all occupants, belted occupants, and unbelted occupants:

Figure 3

Figure 4

Figure 5

The figures show the cumulative distribution of injuries by delta V for fatalities, for MAIS 3+ injuries, and for MAIS 2+ injuries. MAIS 3+ injuries are those which are classified as serious or greater injury, while MAIS 2+ are those which are classified as moderate or greater. (17)

We can see several things by examining the figures. About 50 percent of fatalities in frontal crashes occur at delta V's below 48 km/h (30 mph), and about 50 percent occur at delta V's above 48 km/h (30 mph). Looking separately at unbelted and belted occupants, 51 percent of the fatalities involving unbelted occupants and 47 percent of the fatalities involving belted occupants occur in frontal crashes at delta V's below 48 km/h (30mph). We note that the delta V in NASS represents the speed at which the vehicle would strike a rigid barrier to duplicate the amount of energy absorbed in the crash. Thus, about half of fatalities in frontal crashes occur in crashes that are more severe than a 48 km/h (30 mph) rigid barrier crash, and half of all frontal crash fatalities occur in crashes that are less severe than a 48 km/h (30 mph) rigid barrier crash. Given that Standard No. 208's unbelted crash test requirements are intended to save lives, we disagree that 48 km/h (30 mph) rigid barrier crashes are unrepresentative of the kinds of crashes in which we are seeking to ensure protection.

As to the argument that returning to the unbelted 48 km/h (30 mph) rigid barrier test would prevent continued use of "depowered" air bags and require use of "overly aggressive" air bags, the agency will have to consider the information available to it in making a final decision. (18)

In the NPRM, we noted that, based on very limited data, it appeared that many, perhaps most, vehicles with redesigned air bags continue to meet the historical 48 km/h (30 mph) rigid barrier requirements of Standard No. 208 (using 50th percentile adult male dummies and applying the current injury criteria performance limits) by fairly wide margins. At that time, we had tested five vehicles with redesigned driver air bags in unbelted 48 km/h (30 mph) rigid barrier tests, and all passed Standard No. 208's previous injury criteria by significant margins. We had tested six vehicles with redesigned passenger air bags in unbelted 48 km/h (30 mph) rigid barrier tests, and all but one passed the standard's injury criteria performance limits by significant margins.

Some vehicle manufacturers objected to our analysis in this area. They argued that, given the variability associated with testing different vehicles of the same design, the fact that a particular vehicle had passed a single test would not necessarily allow them to certify that model vehicle as complying with Standard No. 208 because there would not be a sufficient margin of compliance to ensure that all vehicles of that model would pass the test. Some manufacturers indicated that they need a 20 percent margin of compliance in order to so certify. Vehicle manufacturers also stated that they need to ensure that all variations and configurations of a model would pass the test and that, in some cases, we tested a configuration which would result in lower injury criteria readings than other variations and configurations.

We continue to believe that a key way of assessing the validity of the argument that a return to the 48 km/h (30 mph) barrier test would -- at least in the absence of additional technological improvements -- prevent continued use of redesigned air bags is to test vehicles with those air bags in 48 km/h (30 mph) barrier tests and see how they perform. Therefore, since issuing our NPRM, we have conducted more barrier tests of vehicles with redesigned air bags.

We have now tested a total of 13 MY 1998-99 vehicles with redesigned air bags in a perpendicular rigid barrier crash test at 48 km/h (30 mph) with unbelted 50th percentile adult male driver and passenger dummies. The vehicles represented a wide range of vehicle types and sizes. In particular, the 13 vehicles included one sub-compact car, one compact car, four mid-size cars (representing high sales volume vehicles), one full-size car, two mid-size sport utility vehicles, one full-size sport utility vehicle, one pickup truck, one minivan, and one full-size van. (19)

For the driver position, 12 of the 13 vehicles passed all the relevant injury criteria performance limits we are proposing in this SNPRM. In the one vehicle with a failure, the MY 1999 Acura RL, the driver dummy exceeded the femur load criteria. For the passenger position, 12 of the 13 vehicles also passed all of the relevant injury criteria performance limits. The MY 1998 Dodge Neon slightly exceeded the 60 g chest acceleration limit (with a value of 61.4 g). The other proposed injury criteria performance limits, (i.e., for HIC, chest deflection, and Nij) were easily met in all the tests; for most there was a greater than 20 percent margin of compliance for both the driver and passenger.

Thus, the tested vehicles with redesigned air bags, ranging widely in vehicle type and size, appear to continue to meet Standard No. 208's 48 km/h (30 mph) unbelted rigid barrier test requirements for 50th percentile adult male dummies, many of them by wide margins.

As to any vehicles that do not meet that test, at this point we simply note that TEA 21 affords lead time before all vehicles must meet whatever tests are incorporated in the final rule to be issued in this rulemaking.

As to the issue of margin of compliance, we agree that manufacturers need to ensure that all of their vehicles meet a test requirement established by a Federal safety standard. However, we do not agree that this means a 20 percent margin of compliance is necessary. The chest g value is the injury criterion that is most likely to be the limiting factor in certifying to the 48 km/h (30 mph) unbelted rigid barrier test requirements for the 50th percentile adult male dummy. Examination of compliance and certification data for pre-redesigned air bags shows that manufacturers often certified vehicles to the requirement with much less than a 20 percent margin of compliance. In fact, margins of compliance for our 48 km/h (30 mph) tests of vehicles with redesigned air bags were not that different from those with pre-redesigned air bags.

We are not suggesting that every current production vehicle would comply with the unbelted 48 km/h (30 mph) rigid barrier test. Instead, we are pointing out that a wide ranging sample of vehicle types and sizes meet the 48 km/h (30 mph) rigid barrier test, for 50th percentile adult male dummies, with redesigned air bags.

However, the ultimate issue of this rulemaking is not whether some MY 1998-99 vehicles with redesigned, single-inflation level air bags currently would not meet the 48 km/h (30 mph) unbelted barrier test requirement. As noted above, many of the air bags in current vehicles were not comprehensively redesigned, but are merely older designs of air bags with less power. TEA 21 mandates the issuance of a final rule based on means that include advanced air bag technologies. We believe the selection of future compliance tests under TEA 21 must be made in the context of those technologies, and not in the context of today's less sophisticated one-size-fits-all air bag designs. Today's air bag systems are not advanced air bags and thus do not respond to factors such as crash severity, occupant weight and occupant location. By contrast, the incorporation of advanced technologies would make air bag systems responsive to those factors. If a manufacturer decided to use a somewhat more powerful air bag to meet a 48 km/h (30 mph) unbelted rigid barrier test, or to provide protection in more severe crashes, the manufacturer could use advanced air bag technologies to provide less powerful levels of inflation in lower severity crashes, for smaller occupants, for belted occupants, and for occupants sitting with the seat in the full-forward position. Manufacturers could also reduce aggressivity of air bags by various means such as optimizing fold patterns, different cover designs, lighter fabrics, etc. Advanced technologies would also enable the manufacturer to suppress air bag deployment in appropriate circumstances, such as when children are present.

As we assess the type and severity level of an unbelted crash test should be included in Standard No. 208, we recognize that we must bear in mind that the issue of the suitability of a unbelted 48 km/h (30 mph) rigid barrier test cannot be determined solely based on whether manufacturers can meet that test with redesigned air bags using 50th percentile male dummies. In the NPRM, we proposed not only to return to that test requirement, but also to require vehicles to be certified to several new crash test requirements and new injury criteria performance limits, including tests using 5th percentile adult female dummies in the full forward seat track position, and to requirements to minimize air bag risks. Vehicle manufacturers commented that some of the design options that are available in redesigning their air bags involve potential trade-offs in meeting the different proposed requirements. For example, the optimum size air bag for meeting test requirements for 50th percentile adult dummies may make it more difficult to meet requirements for 5th percentile adult female dummies, and vice versa. This issue, and the agency's testing of current vehicles to a variety of the proposed test requirements, are discussed later in this notice.

Proposed alternative unbelted crash tests

In the NPRM, we indicated that while we believe the 48 km/h (30 mph) unbelted rigid barrier test is a good approach, we were also willing to consider alternative unbelted crash tests. The only alternative unbelted crash test advocated by a commenter that could realistically be implemented within the time frame of this rulemaking is the unbelted 56 km/h (35 mph) offset deformable barrier test suggested by IIHS. As noted earlier, IIHS stated that this configuration would address not only protection in asymmetric crashes but also some issues of intrusion that are related to restraint system performance, e.g., steering column movement.

Given the continued debate over what requirements should apply to ensure protection to unbelted occupants, we want to be sure that we have considered and received the benefit of public comments on the various alternative approaches that are available at this time. One approach, of course, is the one we proposed in the NPRM, the unbelted rigid barrier test. We note that some have suggested that, instead of conducting this test at speeds up to 48 km/h (30 mph), we reduce the maximum speed. Ford, for example, suggested in 1995 that we adopt an upper speed of 40 km/h (25 mph). It coupled this suggestion with the further suggestion that the speed of the belted test be increased to 56 km/h (35 mph). (20) In its recent consensus statement, the Alliance has suggested a single speed test (perpendicular impact only) of 40 km/h (25 mph).

A second possible approach is an unbelted fixed offset deformable barrier test, along the lines suggested by IIHS in its comment on the September 1998 NPRM. While, as discussed above, that organization has recently identified some concerns about that test, we believe an unbelted offset deformable barrier test represents a sufficiently interesting alternative approach to warrant seeking public comment. As to the concern that IIHS recently identified about air bag systems possibly having difficulty distinguishing between the offset test and a low speed crash during the time the decision whether to deploy the air bag must be made, we note that it may be possible to address this potential problem by using advanced sensing systems. That is one of the issues for which we would like to receive public comments. By requesting public comments, we will obtain additional data and views to better enable us to make a thorough evaluation of the merits of including such a test in Standard No. 208.

For this SNPRM, we are proposing and seeking comments on two alternative unbelted tests. The first alternative is the unbelted rigid barrier test (perpendicular and up to ± 30 degrees oblique to perpendicular with 50th percentile adult male dummies, but perpendicular only in tests with 5th percentile adult female dummies) with a maximum speed to be established within the range of 40 to 48 km/h (25 to 30 mph). As part of this alternative, we are considering the possibility of coupling a lower speed for the unbelted barrier test with a higher speed for the belted barrier test. The second alternative is an unbelted offset deformable barrier test with a maximum speed to be established within the range of 48 to 56 km/h (30 to 35 mph). A vehicle would have to meet the requirements both in tests with the driver side of the vehicle engaged with the barrier and in tests with the passenger side engaged.

We note that, in considering a range of upper severity levels, the upper severity level could be adjusted by either changing the test speed or applying different injury criteria limits at higher speeds. For example, in our rulemaking to facilitate quick redesign of air bags, in lieu of the sled test, we identified the possibility of maintaining the 48 km/h (30 mph) unbelted rigid barrier test, but relaxing the limit on chest g's. We also note the possibility of specifying relaxed injury criteria performance limits or lower maximum test speeds that would apply during the TEA 21 phase-in period and more stringent ones that would apply thereafter.

For all of the unbelted crash tests proposed in this document, protection would be required in crashes ranging from a specified minimum speed to a specified highest speed, rather than at all speeds "up to" that specified highest speed.

Under the unbelted rigid barrier test alternative, the agency would not test at a speed of less than 29 km/h (18 mph), and under the unbelted offset deformable barrier test alternative, the agency would not test at a speed of less than 35 km/h (22 mph). (We are proposing a higher minimum test speed for the latter alternative because, for a given speed, it is a less severe test.) This is a departure from the proposal in the NPRM and from prior agency practice. One reason for this change is that we want to be sure that the standard does not push deployment thresholds downward, i.e., cause air bags to be deployed at lower speeds than are appropriate for maximum occupant protection. Commenters indicated that, in order to meet neck injury criteria, air bag deployments might be required at very low speeds, even in crashes with a delta-V lower than 10 mph, particularly with the 5th percentile adult female dummy in the full forward position. While the issue of the most appropriate threshold for air bag deployment is complex, we believe there is a consensus that "no fire" thresholds should not be any lower than they are at present. Moreover, neck injuries are not a significant problem in lower speed crashes.

The proposed high speed unbelted offset deformable barrier test would involve the same crash configuration as we proposed in the NPRM for the up-to-40 km/h (25 mph) belted offset deformable barrier test. Vehicles would have to meet the requirements in tests with both the vehicle and the passenger side of the vehicle engaged. The test would, of course, be conducted at higher speeds, and unbelted 50th percentile adult male dummies and 5th percentile adult female dummies would be used.

The offset deformable barrier test is used in several ways in different parts of the world. The test has been adopted as a requirement in Europe at a speed of 56 km/h (35 mph), using belted 50th percentile adult male dummies, pursuant to EU Directive 96/79 EC. The test is also conducted in Europe at a higher speed, 64 km/h (40 mph), as part of the European New Car Assessment Program. The Australian New Car Assessment Program conducts the same test at the same speed. IIHS also conducts this test at the same speed, using belted 50th percentile adult male dummies to evaluate the crashworthiness of vehicles. Transport Canada is developing a test procedure using belted 5th percentile adult female dummies at impact speeds up to 40 km/h (25 mph) to evaluate air bag sensor performance and air bag aggressivity.

While a great deal has been written on the subject of unbelted rigid barrier tests over the years, the high speed unbelted offset deformable barrier test is relatively new. We note that we have been conducting research for several years with the intention of proposing to add a high speed belted frontal offset test to Standard No. 208. For information about this research program, see our Report to Congress, Status Report on Establishing a Federal Motor Vehicle Safety Standard for Frontal Offset Crash Testing, April 1997. This report is available on our web site at http://www.nhtsa.dot.gov/cars/rules/CrashWorthy/offrt.html.

In our Report to Congress, and in the NPRM (63 FR 49958, at 49960), we stated that we were considering adding the European high speed belted frontal offset test to Standard No. 208 as a supplement to the existing tests. We stated in the Report that the Standard No. 208 rigid barrier test is most effective in preventing head and chest injuries and fatalities, but noted that it does not address lower limb and neck injuries.

We stated further in the Report that while the frontal rigid barrier test of Standard No. 208 does not produce the vehicle intrusion observed in many real world crashes, it does depict those impacts which produce the highest risk of serious to fatal injuries resulting from frontal crashes. We stated that the European frontal test procedure does not address the highest risk of serious to fatal injuries occurring in frontal crashes and that, from our viewpoint, the European test conditions were not acceptable as an alternative to Standard No. 208. We stated, however, that adoption of the European test could yield benefits in terms of a reduction in lower limb injuries.

While our analysis of the European test was made in the context of a belted condition, it nonetheless raises the issue of whether the test is adequately representative of potentially fatal crashes. To address this issue, we have sought to compare the 56 km/h (35 mph) offset deformable barrier crash test recommended by IIHS to a 48 km/h (30 mph) rigid barrier test.

Among other things, we have conducted 56 km/h (35 mph) offset deformable barrier crash tests on MY 1999 Dodge Intrepid and Toyota Tacoma vehicles. Comparing the crash pulses for these tests with the pulses of 40 and 48 km/h (25 and 30 mph) rigid barrier tests that we also conducted using these vehicles, we can make several observations. For each vehicle, there is a long duration, low magnitude acceleration during the initial phase of the test that is associated with the crushing of the deformable barrier face. After the crushing of the barrier face, the remaining segment of the crash pulse is similar to that for the 40 and 48 km/h (25 and 30 mph) rigid barrier tests, and this portion of the acceleration profile generally would fall in between the pulses for those two rigid barrier tests if adjusted with a time shift.

A close look at these pulses suggests that, from the perspective of delta-V, the deformable barrier test is approximately equal in severity to a 45 km/h (28 mph) rigid barrier test. This is consistent with a rule of thumb within the research community that the offset test's barrier equivalent velocity is approximately 20 percent less than the impact speed.

This observation is also supported by findings from our Advanced Frontal Research Program. We provided a number of vehicles tested in both collinear and oblique offset tests to NASS investigators for analysis. The investigators estimated delta Vs that were substantially lower than the impact speeds. (21) Also, IIHS conducted a similar study and observed similar results,(22) i.e., the range of delta Vs were 15 to 28 percent lower than the impact speeds.

It is important to note that although we estimate 45 km/h (28 mph) as the rigid barrier equivalent speed for the 56 km/h (35 mph) offset deformable barrier test, this does not mean that air bags designed to meet the 56 km/h (35 mph) offset deformable barrier test would provide a level of protection equivalent to that provided by air bags designed to meet a 45 km/h (28 mph) barrier-like crashes.

When looking at the severity of a crash and its influence on air bag design, delta V is not the only important factor. Another important factor is the time to reach that delta V. The time is important because it affects the speed at which the occupant strikes the interior of the vehicle, i.e., for a given delta V crash, the shorter the time duration, the higher the occupant impact speed.

As discussed in the test procedures paper, the offset crash test has a long duration deceleration pulse. As a result, occupants in a vehicle involved in such a crash would impact the interior components at lower speeds than occupants who were in a vehicle involved in barrier-like crashes. Because of this aspect of offset crashes, the test procedures paper separates the crash events in NASS and estimates a substantially lower target population for the offset test than for the rigid barrier test.

The high speed unbelted rigid barrier test and the high speed unbelted offset deformable barrier test are significantly different, and each has potential advantages as compared to the other.

Among the considerations that are relevant to the high speed unbelted rigid barrier test are the following-

Among the considerations that are relevant to the high speed unbelted offset deformable barrier test are the following:

As noted above, the concept of a high speed unbelted offset deformable barrier test is new, so there are very few available data for this test. However, we have tested two vehicles, the MY 1999 Toyota Tacoma and Dodge Intrepid, in unbelted 56 km/h (35 mph) offset tests using both 50th percentile adult male and 5th percentile adult female test dummies. One vehicle, the Tacoma, was able to meet the proposed injury criteria performance limits without difficulty (for both types of dummies and both left and right impacts), while the other vehicle, the Intrepid, had difficulty, particularly with the Nij injury criteria performance limits. Of course, neither of these vehicles was designed with the offset test in mind, so these tests have little relevance to the issue of whether vehicles could satisfy such a requirement.

Some vehicle manufacturers have expressed concerns about an unbelted high speed offset test. GM has expressed concern about the ability of vehicle sensing systems to be able to sense the soft, deformable barrier face of the offset deformable barrier, and still be able to perform well in real world crashes. According to that company, its review of actual vehicle data traces plotting deceleration over time indicates that the frontal offset barrier impact initially looks much like a low speed crash, where no air bag or just a first stage air bag might be used. Because of this, a sensor system might not recognize until well into the crash that the vehicle is undergoing a higher speed, severe crash. GM believes that if this test were made a part of the standard, manufacturers would either have to design their sensors to fire any time they see a lower speed, soft impact, which would cause more low speed deployments, or design the sensors to optimize for real world crashes and risk failing this performance test in the standard.

Honda expressed concern about the similarity in pulses between the 40 km/h (25 mph) offset deformable barrier and the 56 km/h (35 mph) offset deformable barrier crashes. In an August 26, 1999 comment submitted to the docket, Honda stated that, even though these tests are dissimilar in terms of ultimate severity, the crash pulses looked similar during the initial decision period of up to 30 ms. This in part reflects the fact that the initial phase of the test is measuring the deformation of the soft barrier. According to Honda, the vehicle's analytical system will be unable to discern the crash severity and will not be able to accurately predict what stage to fire, or even whether to fire the air bag in a timely fashion. That company indicated that this may result in poor algorithm design.

For additional analysis of the two alternative unbelted tests, readers are referred to the aforementioned paper and supplement prepared by our Office of Vehicle Safety Research concerning potential test procedures for Standard No. 208 and to the Preliminary Economic Assessment which accompanies this SNPRM.

It is important to note that, whatever unbelted test is included in Standard No. 208, manufacturers will be required under the final rule to certify all of their vehicles to a wide variety of new test requirements, and in a very short period of time. The analysis we presented earlier in this document concerning how many vehicles currently appear to meet the 48 km/h (30 mph) unbelted rigid barrier requirements for 50th percentile adult male dummies was intended to address the allegation that a return to the test would prevent continued use of redesigned air bags and require a return to overly aggressive air bags; it did not represent an analysis of how easy it would be to meet that particular test requirement in the context of the overall set of proposed requirements.

In commenting on the NPRM, vehicle manufacturers indicated that, as