Skip to main content

Full text of "Evaluation methodologies for four federal motor vehicle safety standards : FMVSS 214, Side door strength, FMVSS 215, Exterior protection, FMVSS 301, Fuel system integrity, FMVSS 208, Occupant crash protection"

See other formats


TL 

242 



DOT HS-802 346 



Nb8 



EVALUATION METHODOLOGIES FOR FOUR 
FEDERAL MOTOR VEHICLE SAFETY STANDARDS 



FMVSS 301 : Fuel System Integrity _ jhs 
FMVSS 208: Occupant Crash Protection 

Contract No. D0T-HS-6-01518 
May 1977 
Final Report 



FMVSS 214: Side Door Strength 
FMVSS 215: Exterior Protection 




PREPARED FOR: 



U S. DEPARTMENT OF TRANSPORTATION 

NATIONAL HIGHWAY TRAFFIC SAFETY ADMINISTRATION 

WASHINGTON, D.C. 20590 



Document is available to the public through 
the National Technical Information Service, 
Springfield, Virginia 22161 



Prepared for the Department of Transportation, 

National Highway Traffic Safety Administration, under 
Contract No.: D0T-HS-6-01518 . This document is dis- 

seminated under the sponsorship of the Department of 
Transportation in the interest of information exchange. 
The United States Government assumes no liability for 
the contents or use thereof. 



NOTICE 

The United States Government does not endorse products 
or manufacturers. Trade or manufacturer’s names appear 
herein solely because they are considered essential to 
the object of this report. 



TL- 



1. Report No. 

DOT HS 802 346 



2. Government Accession No. 



3. Recipient's Catalog No. 



4. Title and Subti^l^ 

EVALUATION METHODOLOGIES FOR FOUR 
FEDERAL MOTOR VEHICLE SAFETY STANDARDS, 

FMVSS 214: Side Door Strength FMVSS 301: Fuel System Integrity 

FMVSS 215: Exterior Protection FMVSS 208: Occupant Crash Protection 



5. Report Date 
May 197 7 



6. Performing Organization Code 



7. Author(s) 

Gaylord M. Northrop 



8. Performing Organization Report No. 

CEM Report 4207-568 



9. Performing Organization Name and Address 

The Center for the Environment and Man, Inc. 
275 Windsor Street 
Hartford, Connecticut 06120 



10. Work Unit No. 



11. Contract or Grant No. 

DOT-HS-6-015 18 



12. Sponsoring Agency Name and Address 

U.S. Department of Transportation 

National Highway Traffic Safety Administration 

400 Seventh Street, S.W. 

Washington, D.C. 20590 



13. Type of Report and Period Covered 

Final Report 

Oct. 1976 - March 1977 



14. Sponsoring Agency Code 



Dept of Transput®®? 



15. Supplementory Notes 



16. Abstract 






oUrtlbl#/ 



Library 



1 . 



This is the final report of the study to develop methodologies for evaluating 
the effectiveness of four selected Federal Motor Vehicle Safety Standards — 
FMVSS 21^: Side Door Strength; FMVSS 215: Exterior Protection; FMVSS 301: 

Fuel System Integrity; and FMVSS 208: Occupant Crash Protection. This report 
provides a summary and overview of the nine preceding reports and an integra- 
tion of the individual approaches developed for evaluating each Standard in 
earlier reports. This report includes conclusions and recommendations, reviews 
of the four Standards, approaches to evaluating the Standards, discussion of 
the methodologies for evaluation, and alternative implementation plans for 
performing the evaluations, individually and in an integrated fashion. The pro- 
jected cost of the total evaluation program for the four Standards is between 
$1.7 and $2.n million. With the exception of the evaluation of passive 
restraint systems, the evaluations could be completed within 24 months, but a 
more cost effective program could take three to four years. 



17. Key Words 

Federal Motor Vehicle Safety Standards 
Evaluation Methodologies 
FMVSS 214, FMVSS 215, 

FMVSS 301, FMVSS 208 
Statistical Methods 



18. Distribution Statement 

Document is available to the public through 
the National Technical Information Service, 
Springfield, Virginia 22161 



19. Security Cl assl f . (of this report) 


j 20. Security Class1f.(of this page) 


21. No. of Pages 


22. Price 


UNCLASSIFIED 


UNCLASSIFIED 

1 


147 





i 




11 



EXECUTIVE SUMMARY 



This report summarizes the work performed by The Center for the Environment and 
Man, Inc. (CEM) to design statistical methodologies and implementation plans 
for evaluating the effectiveness of four specified Federal Motor Vehicle Safety 
Standards (FMVSS) . The four Standards that have been examined are: 

• FMVSS 214 - Side Door Strength 

• FMVSS 215 - Exterior Protection (Bumpers) 

• FMVSS 301 - Fuel System Integrity 

• FMVSS 208 - Occupant Crash Protection 

This report includes conclusions and recommendations about evaluating the Stan- 
dards, reviews of the Standards, approaches to their evaluation, discussion of 
the evaluation methodologies, and implementation plans for doing the evaluation, 
individually and in an integrated fashion. 

Judgmentally , the following comments can be made concerning the feasibility of 
demonstrating the effectiveness of each of the Standards. Presentation is or- 
dered by greatest likelihood of success in establishing that the Standard meets 
its objectives. 

• FMVSS 208 - Occupant Crash Protection 

- Previous analyses have shown that lap belts and lap/shoulder 

belts are effective in reducing bodily injury in crashes. 

— The analysis proposed herein will sharpen the results 
of previous studies and attempt to include the effect 
of crash speed and direction. 

- Preliminary review of tests involving passive restraint systems, 

such as cited in the Secretary's June 9, 1976 statement, have 
suggested they are effective in reducing bodily injury in 
crashes . 

— The analyses proposed herein will provide the estimate (s) 
of effectiveness, but adequate data for passive systems 
will probably not be available for at least three years. 

• FMVSS 215 - Exterior Protection 

- Fragmented analyses indicated that in low speed front/rear 

crashes the 5 mph bumpers reduce damage to certain vehicle 
parts. (Repair cost may be higher in high speed crashes, but 
that is not involved in the objective of the Standard.) 

— The analyses proposed herein, when considered together, 
will probably be sufficient to determine some aspects 
of the effectiveness of this Standard. 



iii 



• FMVSS 214 - Side Door Strength 

- Existing and anticipated data bases (state mass accident data 

and NCSS data) are likely to be inadequate in terms of injury 
information or number of cases to show the effectiveness of 
side door beams to reduce passenger compartment intrusion 
and occupant bodily injury, with a satisfactory level of 
statistical significance. 

— If additional NCSS-type data are obtained, it is possi- 
ble that the effectiveness of this Standard may be de- 
termined, at least in terms of passenger compartment 
intrusion. The added stiffness due to the side door 
beam may cause a shift in bodily injury from torso to 
head, complicating the analysis of the effectiveness of 
side door beams in reducing bodily injury. 

• FMVSS 301 - Fuel System Integrity 

- We found no existing data readily accessible to determine the 

effectiveness of this Standard. Fuel spillage is not repor- 
ted in accident reports; fire is not (or not unambiguously) 
reported. 

— There appears to be a moderate possibility of determin- 
ing some aspects of the effectiveness of this Standard 
by (1) analyzing frequency of fires and fuel spillage 
from fire department data; (2) frequency of fire- 
related fatalities in automobile accidents; and (3) 
conducting a detailed survey of fuel system rupture in 
towaway accidents. It will probably be necessary to 
conduct all three of these investigations to obtain 
supportive corroboration among results. 

The crucial element in evaluating all the Standards is the availability of suf- 
ficient data which describe all factors with an appreciable influence on the out- 
come of an accident. The second critical problem is that a "model" has to be 
used to separate the effect of the Standard from those of all the other criti- 
cal factors. The types of data bases we considered were: 

• Available automated data bases, such as state accident data tapes, 

the RSEP data base, the NCSS data base (available in early 1978), 
etc. 

• Available data sources from which automated data bases could be 

readily constructed, such as data from fire departments on auto- 
mobile fires and fuel spillage in accidents. 

• New data collection efforts, such as data to essentially augment 

NCSS, mail surveys, special supplementary data to be collected by 
police when preparing standard automobile accident reports. 

The "models" proposed for this analysis are not physical models, based on known 
theoretical or empirical relations. Rather, they are mathematical structures 
which are, in our opinion, sufficiently flexible to adequately describe the re- 
lations to be expected. 



iv 



Various statistical techniques are proposed, primarily dependent on whether 
the data are continuous or whether all or some of them are categorical. The 
final selection, however, will be influenced by the characteristics of the 
actual data available, and by the investigator's preference for and experience 
with specific methods. 

In addition to specifying methods to determine the effectiveness of the four 
Standards, procedures were outlined for selecting vehicle manufacturers, makes, 
models, etc., for a basis for analyzing the direct costs of meeting the Stan- 
dards. Appropriate parts lists were also given. 

To evaluate FMVSS 214 (Side Door Strength ) we recommend that state mass acci- 
dent data be analyzed to determine the effects of vehicle age on intrusion 
and injury, and also to delineate the effects of the gradual implementation of 
side door beams over the years 1969 through 1972. This analysis is secondary 
in importance to the detailed analysis. This information would be used to 
guide the more complex analysis of NCSS data, following its availability after 
March 1978. We expect that there will not be enough side impact cases in the 
NCSS data base to permit determination of effectiveness with regard to reduc- 
tion of intrusion and injury severity, with an acceptable level of statistical 
significance. This initial analysis of NCSS data will provide an opportunity 
to develop and check out the statistical methodology and determine the amount 
of additional data to be collected. The critical element in this evaluation 
is whether the statistical models proposed will control for the complex inter- 
action of factors in side collisions. 

The evaluation of FMVSS 215 (Exterior Protection ) is complicated by the fact 
that there is a lack of detailed da£a on low speed accidents in which there is 
little or no damage. We propose to get certain information from existing State 
Farm Insurance data and possibly from state mass accident data. We recommend 
a mail survey of car owners to get information on the frequency of low speed 
front/rear crashes, and we recommend that towtruck operators be used to collect 
information on the characteristics of vehicles involved in front/rear towaway 
accidents. No single data source is considered adequate to achieve the evalu- 
ation, but it is likely that evaluation will be possible if the several analy- 
ses are performed and used to reinforce each other. An analysis of HLDI data 
is discussed but- because HLDI data have only total claim payment amounts and 
no information on type of crash and many other factors, one cannot expect much 
information will result. 

There are very few data readily available for the evaluation of FMVSS 301 (Fuel 
System Integrity ) . To get information on fire-related fatalities, a number of 
sources would be used to build an analysis data base — FARS data, state mass 
accident files, state fatal accident files, state medical examiner's files, etc. 
We recommend that fire and police department records be used to determine the 
frequency of fire and fuel spillage in accidents. If new data were desired, 
cooperating police departments would be requested to obtain these data for fire/ 
spillage cases, on special forms, while they prepare normal accident reports. 

We also recommend that a detailed data collection effort be undertaken concern- 
ing fuel system rupture in towaway accidents. 



v 



The evaluation of FMVSS 208 (Occupant Crash Protection) builds on the results 
of earlier studies, with regard to the effectiveness of lap and lap/shoulder 
belts. For determining belt effectiveness, we propose analysis of the combined 
NCSS/RSEP data base, after additions to the RSEP data have been completed, so 
that the effect of impact speed and possibly impact direction can be tested. 

For the passive system evaluations, we anticipate that new data will be obtained 
using accident "tracking" methods such as those presently performed by NHTSA, 
Volkswagen, etc. The critical problems in the evaluation will be the delay in 
getting sufficient data on passive restraint vehicle crashes. 

With regard to the implementation of our suggested approaches, we conclude that 
it may take about $2 million and 1.5 to 4 years to perform the effectiveness 
evaluations of these four Standards. However, many more economical and less 
time-consuming programs of evaluation are possible. 

If an integrated program approach is adopted, then we estimate that savings of 
about 14 percent could be achieved, assuming all work is performed by outside 
contractors . 

In our cost estimates, CEM has taken a somewhat conservative position in terms 
of volume and type of new data to be acquired, based on the estimated needs to 
achieve "acceptable" levels of statistical significance. Once the preliminary 
analyses are actually performed, it may be determined that some data are not 
required (or are being obtained as part of other programs) and that certain 
analyses need not be performed. 

Assuming that all work is contracted out, the costs to evaluate the Standards 
are shown in the table below. The table on the next page shows the character- 
istics of alternative implementation programs. 



COSTS OF ALTERNATIVE IMPLEMENTATION PROGRAMS 



FMVSS 


Non-Integrated 

Program 


Integrated 

Program 


Uniform Cost 
Program 


214 

Side Door 
Strength 


$ 479,000 


$ 380,000 
(21*)* 


$ 479,000 


215 

Exterior 

Protection 


$ 335,000 


$ 295,000 
(12*) 


$ 335,000 


301 

Fuel System 
Integrity 


$ 593,000 


$ 470,000 
(21*) 


$ 593,000 


208 

Occupant Crash 
Protection 


$ 601 ,000 


$ 580,000 
(3.5S) 


$ 601 ,000 


Total 


$2,008,000 


$1 ,725,000 


$2,008,000 



‘Percent reduction, relative to the costs for the Non- Integrated 



Program. 



vi 



CHARACTERISTICS OF ALTERNATIVE IMPLEMENTATION PROGRAMS 



Item 


Non-Integrated 

Program 


Integrated 

Program 


Uniform Cost 
Program 











Criteria 



• Each Standard is 

evaluated totally 
independently of the 
others. 

• Evaluation of all 

Standards begins at 
same time, and is 
completed as quickly 
as feasible. 



• Common data bases are 

evaluated for all 
Standards . 

• Available data bases 

are analyzed first. 

• Results of analyses 

are used to form 
base for next phase, 



• Emphasis on equal- 

izing annual funding, 

• Each Standard is 

evaluted totally 
independently of the 
others. 



Cost ($ 000) 

Year 1 
Year 2 
Year 2 
Year 4 



Total 



$ 1,404 
391 
161 
47_ 

$ 2,008 



$ 5 : 6 

616 
347 
176 

$ 1,725 



$ 608 
619 
657 
124 

$ 2,008 



Evaluation 

Schedule 



214 

215 
301 
208 



Year 



Year 



Year 



i 




1 


i 

2 









vii 




TABLE OF CONTENTS 



Section 



Title 



1.0 INTRODUCTION 

1.1 Background 

1.2 Objectives 

1.3 Scope 

1.4 Approach 

1.5 Limitations 

1.6 Outline of the Report 

2.0 CONCLUSIONS AND RECOMMENDATIONS 

2.1 Conclusions 

2.2 Recommendations 

3.0 REVIEW OF STANDARDS 

3.1 Review of FMVSS 214 - Side Door Strength 

3.2 Review of FMVSS 215 - Exterior Protection 

3.3 Review of FMVSS 301 - Fuel System Integrity 

3.4 Review of FMVSS 208 - Occupant Crash Protection 

3.5 References for Section 3.0 

4.0 APPROACHES TO EVALUATING THE STANDARDS 

4.1 Approaches for Evaluating Individual Standards 

4.2 Integrating the Evaluation Approaches 

4.3 Cost Data 

5.0 METHODOLOGIES FOR EVALUATING THE STANDARDS 

5.1 Introduction 

5.2 Sources of Data 

5.3 Statistical Techniques 

5.4 Hardware Cost Data Acquisition 

5.5 References for Section 5.0 

6.0 IMPLEMENTATION PLAN 

6.1 Introduction 

6.2 Early Results, Non- Integrated Plan 

6.3 Integrated, Reduced Cost Plan 

6.4 Early Results and Equalized Funding Plan 

7.0 END PRODUCTS OF THIS STUDY 



APPENDIX A: 
APPENDIX B: 
APPENDIX C: 
APPENDIX D: 



FEDERAL MOTOR 
FEDERAL MOTOR 
FEDERAL MOTOR 
FEDERAL MOTOR 



VEHICLE SAFETY 
VEHICLE SAFETY 
VEHICLE SAFETY 
VEHICLE SAFETY 



STANDARD NO. 214 
STANDARD NO. 215 
STANDARD NO. 301 
STANDARD NO. 208 



Page 

1-1 

1-1 

1-1 

1-1 

1-2 

1-2 

1-2 

2-1 

2-1 

2-8 



3-1 

3-1 

3-5 

3-9 

3-14 

3- 19 

4- 1 

4-2 

4-10 

4- 10 

5- 1 

5-1 

5-1 

5-7 

5-21 

5- 27 

6 - 1 

6-1 

6-6 

6-8 

6-10 

7-1 



viii 



APPENDIX E; DISCUSSION OF STATISTICAL METHODS 



APPENDIX Fi INTRODUCTION DATES FOR SIDE DOOR REINFORCEMENT BEAMS 

APPENDIX G; COST ESTIMATING METHODOLOGIES USED BY BLS, GAO, AND NHTSA 

APPENDIX H. A STATISTICAL METHOD FOR COST DATA ACQUISITION: 

HOW TO SELECT THE MAKE AND MODEL PRODUCED BY A MANUFACTURER 

APPENDIX I: FLOWCHARTS OF SELECTED ANALYSES 



ACKNOWLEDGMENTS 



The following CEM staff made major technical contributions to this study: 
Mr. John Ball, Mr. Gary Haas, Dr. Leonard Hafetz, Dr. Hans Joksch, and Mr. 
Joseph Reidy. Dr. Gaylord Northrop was Project Manager. Ms. Kayla Costenoble 
provided the graphics and considerable editorial input for all ten reports. 

Mrs. Teresa Mayer and Mrs. Carmela Miller typed the many versions of the pre- 
liminary and final study designs, and Ms. Marjorie Wallace was responsible for 
reproduction. 

The CEM study team is also indebted to the contributions of the consultants 
Drs. Uwe Koehn and Alan Gelfand of the University of Connecticut, and Mr. Robert 
Cromwell, P.E. 

In addition, Mr. Warren LaHeist of the Office of Program Evaluation and 
the Contract Technical Monitor provided the study team with useful feedback on 
our progress. We would also like to thank Mr. Nelson Gordy and the NHTSA spec- 
ialists for the four Federal Motor Vehicle Safety Standards and other NHTSA 
staff who provided assistance and reviewed the reports. 



x 



ABBREVIATIONS USED 



ACRS 


Air-Cushion Restraint System 


AIS 


Abbreviated Injury Scale 


AMC 


American Motors Corporation 


ANACOVA 


Analysis of Covariance 


BEV 


Barrier Equiavlent Velocity 


BLS 


Bureau of Labor Statistics 


CALSPAN 


Formerly Cornell Aeronautical Laboratories 


CDC 


Collision Deformation Classification 


DOT 


Department of Transportation 


FARS 


Fatal Accident Reporting System 


FMVSS 


Federal Motor Vehicle Safety Standard 


GAO 


General Accounting Office 


GM 


General Motors 


GVWR 


Gross Vehicle Weight Range 


HLDI 


Highway Loss Data Institute 


HSRI 


Highway Safety Research Institute 


MFV 


Multipurpose Vehicles 


MVP 


Motor Vehicle Programs 


NASS 


National Accident Sampling System 


NCSS 


National Crash Severity Study 


NHTSA 


National Highway Traffic Safety Administration 


OIC 


Occupant Injury Classification 


RSEP 


Restraint System Evaluation Program 


SWRI 


Southwest Research Institute 


TAD 


Traffic Accident Data Project Scale 


use 


University of Southern California 


VW 


Volkswagen 



xi 



1 . 0 INTRODUCTION 



1.1 Background 

The first Federal Motor Vehicle Safety Standards were issued by the 
National Highway Traffic Safety Administration in 1967 and 1968 for 1968 and 
1969 model cars. An essential problem with these and subsequent Standards is 
to determine whether they are effective in achieving the purpose for which they 
were enacted. 

This study was one of two independent studies funded by NHTSA's Office of 
Program Evaluation to develop methodologies to evaluate four Federal Motor 
Vehicle Safety Standards. The Standards selected for study were: 

• FMVSS 214 - Side Door Strength 

• FMVSS 215 - Exterior Protection 

0 FMVSS 301 - Fuel System Integrity 

• FMVSS 208 - Occupant Crash Protection. 

The Center for the Environment and Man, Inc. (CEM) completed this study in six 
months, producing ten reports and two briefings for NHTSA. (See Section 7.0 
for a list of end products of this study.) 

1.2 Objectives 

The overall objectives of the study were to develop methodologies to eval- 
uate the four FMVSS. The specific objectives to achieve the overall goal were to 

• Review background material on the four Standards. 

• Study the feasibility of evaluating the effects of each of 

the four Standards. 

• Develop a study design which would provide estimates of effects 

of a Standard given certain confidence limits and sample sizes. 

• Prepare a detailed work plan to implement the study design. 

• Describe in detail the procedures for processing the data and 

performing the evaluations. 



1.3 Scope 

The study was limited to six months, during which the study was broken up 
into four phases. The first phase was one month long and satisfied the first 
specific study objective — review background material. The second phase covered 
the next two months and the next two specific study objectives — feasibility and 
preliminary design of an evaluation procedure. A report was prepared for each 
of the four Standards. The third phase covered the next two and one-half months 
and addressed the final two specific obj ectives— final design and implementation 
plan for evaluating the effectiveness of the Standard. Four reports were pre- 
pared. The last phase covered the final half month of the study and focused 
on integrating the results of the previous nine reports and preparing the final 
report . 



1-1 



1.4 Approach 



Our overall approach was to try to develop methods which would utilize 
existing data to provide some preliminary information on the effects of the 
Standard and to guide the collection and analysis of new data. The approach 
taken by CEM in developing the preliminary study designs involved intensive 
interaction between study team members. Special meetings between project staff 
and statistical consultants on the nature of existing and potential data 
evolved toward specific analytic tools — regression models with analysis of co- 
variance, log-linear models, contingency table analysis, log-normal distribu- 
tions, etc. After the preliminary study designs were developed, CEM refined 
them for actual implementation. Finally, after the final design and imple- 
mentation plans for the individual Standards were finished, an effort was made 
to integrate the separate plans » and three alternative programs were developed. 



1.5 Limitations 

The task of developing a detailed plan for performing a complex statisti- 
cal analysis of data is extremely difficult to do in the abstract. Many de- 
cisions are determined by the nature of the data and, in this case, actual 
testing of our proposed methods was precluded. 

Secondly, some material was generated during the study which does not 
directly serve to evaluate the effectiveness of a Standard, but was desirable 
from the point of view of background. These are such items as the general dis- 
cussion of statistical methods, the discussion of cost estimating methodologies, 
etc. In addition, some items were outlined in more detail for comprehensiveness, 
but they do not directly address the question of effectiveness. These are (1) 
the analysis of HLDI claim payment data becuase of the aggregation of all acci- 
dents, the dollar amounts, and the biased nature of the information, and (2) 
the restraint system usage survey, which would only provide information on the 
differences between usage in the general driver populaion Vs. the accident 
population. 

1.6 Outline of the Report 

Section 2 presents conclusions and recommendations. Section 3 reviews the 
Standards. Section 4 discusses the approaches to evaluating the Standards. 
Section 5 deals with the specific methodologies which are suggested to analyze 
the Standards. Section 6 presents individual and integrated implementation 
plans. Section 7 lists the end products generated during this study. 

The appendices contain copies of the latest version of the four Standards, 
a general discussion of statistical methods, some specific discussion of NHTSA, 
General Accounting Office, and Bureau of Labor Statistics costing methodologies, 
and other supporting information. 



1-2 



2.0 CONCLUSIONS AND RECOMMENDATIONS 



2.1 Conclusions 

We conclude that it may take about $2 million and one and a half to four 
years to perform the effectiveness evaluation of the four Standards. Gener- 
ally, we feel that the likelihood of successfully estimating the effectiveness 
of the Standards are, in order: 

• FMVSS 208 - Occupant Crash Protection 

- Previous studies have shown the effectiveness of lap and 

lap/shoulder belts. The suggested analysis will extend 
previous research to include the effect of impact speed 
and direction. 

- The effectiveness of passive restraint systems has been 

demonstrated in test situations. The suggested analysis 
will establish their effectiveness under field conditions 
on a large scale. 

• FMVSS 215 - Exterior Protection 

- Given that tests have demonstrated the effectiveness of 

the 5 mph bumpers under certain conditions, the proposed 
analyses, when considered together, will probably be 
sufficient to reveal its effect in real accident conditions. 

• FMVSS 214 - Side Door Strength 

- Existing data bases are likely to be inadequate to 

delineate the effectiveness of side door beams with a 
satisfactory level of confidence. The collection of 
additional detailed data which is targeted for 
specific categories may provide a sufficient data sample 
size to estimate the effectiveness. 

• EHVSS 301 - Fuel System Integrity 

- We know of no existing data which are readily accessible to 

determine the effectiveness of this Standard. The fre- 
quency of vehicle fire or fuel spillage due to accidents 
is low. Special data collection would be needed to evaluate 
the Standard. 

Table 2-1 below gives a complete overview of our conclusions on how the 
Standards should be evaluated. 



2-1 



TABLE 2-1: SUMMARY OF CONCLUSIONS 



Items 


FMVSS 214: 

Side Door Strength 


FMVSS 215: 
Exterior Protection 


Recommended 

Approach 


• Perform detailed statistical analyses 

of NCSS data to determine: 

- Initial estimates of effective- 

ness. 

- Significance of initial estimates. 

- Need for additional new data, 

if any. 

• Collect additional data, as necessary, 

to achieve desired levels of signifi- 
cance of results, and repeat the 
detailed statistical analyses. 

• Conduct auxiliary analysis of 

existing mass accident data to 
determine: 

- Vehicle age effects. 

- Effects of gradual implementation 

of side beams in 1969-1972 
model year cars. 

• Collect and analyze direct costs of 

side door beam hardware required to 
meet the Standard, using a statis- 
tical sampling method. 


• Analyze existing data: 

- State Farm Mutual Insurance Company 

— Use auto accident claim data 
to determine the frequency 
of bumper-related part damage. 

- Mass accident data 

— Determine if over time there 
has been a shift in vehicle 
damage away from bumper areas. 

- Highway Loss Data Institute (HLDI ) 

— Determine if there has been a 
shift in average claim pay- 
ments, over time, due to the 
Standard. 

• Collect and analyze new data: 

- Car owner survey 

-- Determine the difference in 
frequency of no-damage, unre- 
ported damage low speed acci- 
dents for pre-and pcst-Stand- 
ard cars. 

- Towaway survey 

— Collect data from tow truck op- 
erators on the frequency of 
towing in front/rear accidents. 

• Collect and analyze direct costs of 

bumper-related hardware required to 

meet the Standard, using a statisti- 
cal sampl i ng method. 


Measures of 
Effectiveness 


• Reduction in intrusion due to side 

impact. 

• Reduction in injury severity. 

c Shift in bodily injury location. 


• Reduction of frequency of damage to 

safety-related and bumper-related 
parts. 

• Reduction in car accident claim 

payments. 

• Reduction of towing frequency in 

front/rear accidents. 



2-2 



FMV5S 301: 

Fuel System Inteqrlty 


FMVSS 208: 

Occupant Crash Protection 


• Collect and analyze data on fuel sys- 

tem rupture In towaway accidents 
(new data collections). 

• Analyze the frequency of fire/fuel 

spillage accidents using existing 
fire/police department data (or 
possibly data newly-collected by 
police agencies). 

• Analyze the frequency of fire- 

related motor vehicle fatalities 
using data from Fatal Accident 
Reporting System (FARS) and state 
fatal accidents files. 

• Collect and analyze direct costs of 

fuel system hardware required to 
meet the Standard, using a statis- 
tical sampling method. 


• Use NCSS and RSEP data bases to 

analyze the effect of impact speed 
and (possibly) direction on the effec- 
tiveness of lap and lap-shoulder 
belts. 

• Use existing and new accident data on 

vehicles equipped with passive re- 
straint devices to evaluate their ef- 
fectiveness. Perform the analysis In 
stages as significant data are col- 
lected by the tracking program. 

'• Conduct a seat belt usage survey to 
allow determination of restraint sys- 
tem use for the entire car driving 
population, 

• Collect and analyze direct costs of 

restraint system hardware required 
to meet the Standard, using a statis- 
tical sampling method. 


• Reduction of frequency of fuel 

system rupture in towaway acci- 
dents. 

• Reduction of frequency of fire or 

fuel spillage In all accidents. 

• Reduction of fire-related motor 

vehicle fatalities. 


• Reduction in Injury severity. 



2-3 



TABLE 2-1: SUMMARY OF CONCLUSIONS (Cont.) 



Items 


FMVSS 214: 

Side Door Strength 


FMVSS 215: 
Exterior Protection 


Availability 
of Data to 
Estimate 
Measures of 
Effectiveness 


t Available data bases: 


0 Available data bases: 


- Mass accident data 

-- Texas 

-- North Carolina 

- NCSS 

# New data collection: 

- Needed to supplement NCSS data, 

if level of statistical signi- 
ficance of results obtainable 
with NCSS data is too low. 

- Possibly need more detailed infor- 

mation on passenger compartment 
intrusion than is available in 
NCSS. 


- Mass accident data 

— Texas 
— New York 
— North Carolina 
-- Others 

- State Farm repair and replacement 

data. 

- HLDI claim payments data. 

0 New data collection: 

- Car owner survey of low speed 

accidents. 

- Towaway survey to determine fre- 

quency of towing in front/rear 
accidents. 


Statistical 
Analysis 
Methods 
to be Used 


0 Mass accident data: 


0 Mass accident data : 


- Contingency table analysis. 

0 NCSS data: 

- Log-linear model, with Chi-Square 

goodness-of-fit analysis (all 
categorical variables). 

- Regression analysis with analysis 

of covariance models (some con- 
tinuous and some categorical 
variables). 

- Descriptive index method used to 

delineate effectiveness and pro- 
vide a basis of comparison of re- 
sults from the two methods. 

0 Hardware cost data: 

- Latin square experimental design 

to analyze manufacturer, market 
class, body type stratifica- 
tions. 


- Contingency table analysis. 

0 State Farm repair and replacement data: 

- Contingency table analysis. 

0 Car owner survey data: 

- Contingency table analysis. 

0 Towaway survey data: 

- Contingency table analysis. 

0 HLDI data: 

- Comparison of distribution of pre- 

and post-Standard car payment 
claims, using truncated loq- 
normal distribution theory. 

0 Hardware cost data: 

- Experimental design with two 

replications to analyze manu- 
facturer, market class 
stratifications. 



2-4 



FMVSS 301 : 

Fuel System Inteqrlty 


FMVSS 208: 

Occupant Crash Protection 


• Available data bases: 


e Available data bases: 


- Fatal Accident Reporting System 
(FARS) 

• Data sources for development of 


- NCSS 

- RSEP 

• Data sources for development of 


data bases: 

- State mass accident files and other 

state fatality files. 

- Fire department records on vehicle 

fires and fuel spillage. 

• New data collection: 

- Frequency of fuel system rupture 

in towaway accidents. 

- (Possibly) new data on fire and 

fuel spillage collected by police 
in vehicle accident investigations. 


data bases: 

- Tracking programs for passive re- 

straint system vehicles. 

— NHTSA 

— Allstate Insurance 
-- General Motors 
-- Volkswagen 

® New data collection: 

- Restraint system usage survey. 

- Additional data from tracking pro- 

grams for passive restraint 
vehicles. 


• Fuel system ruDture data: 


9 RSEP/NCSS data: 


- Contingency table analysis for ve- 

hicles with no observable aging 
effects. 

- Trend analysis to determine: 

— Aging effects. 

-- Occurrence of rupture where 
aging effects are discerned. 

• Fire and fuel spillaqe data: 

- Contingency table analysis. 

- Likelihood ratio test. 

• Fire-related fatality data: 

- Contingency table analysis. 

- Likelihood ratio test. 

• Hardware cost data: 


- Log-linear model, with Chi- 

squared goodness-of-fi t analysis 
(all categorical variables). 

- Regression analysis with analysis 

of covariance models (some con- 
tinuous and some categorical 
variables). 

- Descriptive index method use to 

delineate effectiveness and pro- 
vide a basis of comparison of re- | 
suits from the two methods. 

e Passive restraint system data: 

- Same as above. 

• Restraint system usaqe survey data: 


- Experimental design with two 
replications to analyze manu- 
facturer, market class 
stratifications. 


- Tabulations. 

- Estimates of standard errors. 

# Hardware cost data: 1 

- Lap and lap/shoulder belts 

—Balanced incomplete block design 
to analyze manufacturer, seat 
configuration. Inertia reel 
stratifications. 

- Passive systems 

—Consult General Motors and 
Volkswagen. 



2-5 



TABLE 2-1: SUMMARY. OF CONCLUSIONS (Cont.) 



Items 


FMVSS 214: 

Side Door Strength 


FMVSS 215: 
Exterior Protection 




Resources 

Required 

(Special 

Needs) 


• Statistical/computer modeling 

capabilities. 

• Detailed accident investigation 

capabilities. 


t Data processing capabilities. 
• Survey experience. 


Costs 






• Non-Inteqrated Plan 






- Total Cost 
($ 2,008,000) 


$ 479,000 


$ 335,000 


- Person-Years 


9.0 


4.6 


- Computer Costs 


$ 19,000 


$ 10,000 


- Other Costs 


$ 10,000 


$ 95,000 


- Duration (months) 


36 


16 


® Integrated Plan 






- Total Cost 
($ 1 ,725,000) 


$ 380,000 


$ 295,000 


- Duration (months) 


45 


40 


• Time Equalized 
Funding Plan 






- Total Cost 
($ 2,008,000) 


$ 479,000 


$ 335,000 


- Duration (months) 


24 


16 



2-6 



FMVSS 301 : 


FMVSS 208: 


Fuel System Inteqrity 


Occupant Crash Protection 


• Technical field data 
collection capabilities. 


• Statistical/computer modeling 
capabilities. 


• Experience in hard copy 
information retrieval. 


• Survey experience. 


$ 593,000 


$ 601 ,000 


11.0 


10.5 


$ 10,000 


$ 10,000 


$ 33,000 


$ 66,000 


18 


48 


$ 470,000 


$ 580,000 


42 


48 


$ 593,000 


$ 601 ,000 


18 


24 



2-7 



T 



T 



2.2 Recommendations 

It is not possible for CEM to make an unqualified, unique recommendation 
concerning the implementation plan to be followed for evaluating the effective- 
ness and hardware costs of the four Standards considered in this study. This 
is primarily due to the potential interactive effects which data collection 
efforts and results obtained in the Standards evaluation program could have 
with other research and data collection programs currently being conducted or 
planned by NHTSA. CEM is not privy to NHTSA’s plans for the next several years 
in traffic safety research and data collection programs and, hence, cannot 
judge what would be an optimum interface between the Standards evaluation pro- 
gram and other studies. 

With full consideration of the above statements, the following qualified 
recommendations can be made. 



The Integrated Flan is recommended if one is concerned with maximizing 
the interactive relationships among tasks and capitalizing on commonality of 
features concerning data bases, collection efforts and analysis approaches. 
This implementation plan permits cost savings and schedules tasks according to 
certain logical premises. The majority of tasks scheduled during the first 
year require only existing data. Most tasks which depend upon new data col- 
lection or extensive data acquisition are scheduled to start in the second or 
third year. Work proceeds on all Standards throughout the entire four years 
of the project. While intermediate results are available at various times 
during the first three years of the project, final definitive results on the 
evaluation of each of the Standards are not available until the fourth year of 
the project. 



The Time Equalized Funding Plan is recommended if one is concerned with 
obtaining definitive final results on some Standards during the first two years 
and at the same time equalizing the funding level over the first three years 
of the project. The final evaluation results on FMVSS 214 and FMVSS 215 are 
obtained within the first two years, but during this time no work at all is 
carried out on the FMVSS 208 evaluation and the evaluation of FMVSS 301 is not 
started until the second year. The work concentration by year and Standard is: 



o Year 1 
® Year 2 
® Year 3 
© Year 4 



FMVSS 215 and FMVSS 214 
FMVSS 214 and FMVSS 301 
FMVSS 301 and FMVSS 208 
FMVSS 208. 



The Non-Integrated Plan or minor variations of this plan might be desir- 
able if one wants to obtain as many intermediate and final results on the 
evaluation of the four Standards as quickly as possible and if one is willing 
to budget a highly skewed distribution of the funding — with the major portion 
of funds being expended in the first one to two years. This implementation 
plan minimizes time-sequencing of tasks and, hence, does not permit much inter- 
active use of results and analyses among tasks. 



2-8 



3.0 REVIEW OF STANDARDS 



This section reviews and summarizes the essential background information 
which must be considered in developing a plan to evaluate the effectiveness of 
each of four selected Federal Motor Vehicle Safety Standards (FMVSS) . The 
four selected FMVSS which have been examined are: 

• FMVSS 214 - Side Door Strength 

9 FMVSS 215 - Exterior Protection 

• FMVSS 301 - Fuel System Integrity 

• FMVSS 208 - Occupant Crash Protection 

Each Standard is reviewed in a separate subsection in the above-listed order. 

3.1 Review of FMVSS 214 - Side Door Strength 

The rationale for issuing this Standard was the observation that occupant 
injury severity in side-door impact crashes increased with the depth of intrusion. 
To reduce this intrusion, and thereby injury severity, strengthening side doors 
was suggested. Beginning with the 1969 model year, many car models were equip- 
ped with side door guard beams. The Standard became effective on January 1, 

1973, and has not been amended since then. 

Purpose of FMVSS 214 

• The specific purpose is to set strength requirements for side doors. 

• The general purpose is to minimize the safety hazard caused by in- 

trusion into the passenger compartment in a side impact accident. 

General Requirements of FMVSS 214 

Any passenger car side door that can be used for occupant egress must 
meet three crush resistance tests, using a specified test device: 

• Initial Crush Resistance of not less than 2,250 lb. 

• Intermediate Crush Resistance of not less than 3,500 lb. 

• Peak Crush Resistance of not less than 7,000 lb, or two times the 

curb weight of the vehicle, whichever is less. 

Measures of Effectiveness 

The specifications of the Standard are given in terms of a static test. 
Conceptual measures of its real world performance are the intrusions occurring 
in actual crashes, resulting from the dynamic interaction of two vehicles, or 
a vehicle with an object. Conceptual measures of its ultimate effectiveness are 
the expected injury severity in a side door impact crash, or the probability of 
an injury's exceeding a certain level of severity. Both intrusion and injury 
severity are dependent on many pre-crash and crash phase factors. Therefore, 
it appears conceptually impossible to directly evaluate the effect of reduced 
intrusion upon injury reduction. 



3-1 



The ultimate performance measure of FMVSS 214 is its effect on occupant 
injury. To do an adequate statistical analysis of this effect, a specific 
quantitative measure of injury must be available. Unless such a reliable 
measure is available, detecting shifts in injury severity resulting from the 
imposition of FMVSS 214 will be nearly impossible. The requirement for a 
reliable injury severity measure could be relaxed only if the primary effect 
of the Standard was a shift in injury severity at the highest end of the 
scale (e.g., from fatal to seriously injured or from seriously injured to 
minor). Since such a shift is not expected to occur, a comprehensive injury 
scale is necessary. 

Most existing accident data bases rely on police accident reports for 
determination of injury severity. This usually consists of a five point 
scale of K, A, B, C, 0, where: 

® K = Killed 

« A = Serious visible injury 

• B = Minor visible injury 

• C = No visible injury 

• 0 = No injury. 

Though these injury levels are defined more precisely than indicated, 
definitions may vary between jurisdictions, and have changed over time. The 
greatest practical drawback of this scale is that the assignment is made at 
the scene of an accident by a police officer, on the basis of only a few 
visible indications. The greatest conceptual problem is that the "A" cate- 
gory tends to cover a very wide range of injury severity; in effect, it covers 
the entire range of injuries which are of primary concern for evaluating FMVSS 
214. A more satisfactory scale is the Abbreviated Injury Scale (AIS) , which 
is available in some comprehensive data bases (NASS, NCSS)*. It is a seven 
point scale, 0 through 6, where: 

• 0 = No injury 

• 1 = Minor 

• 2 = Moderate 

• 3 = Severe (not life-threatening) 

• 4 = Serious (life-threatening, survival probable) 

• 5 = Critical (survival uncertain) 

• 6 = Maximum (currently untreatable) 

The AIS is precisely defined by a dictionary defining specific injuries for 
six body regions. In the case of multiple injuries, medical judgment is used 
to assign an overall AIS level. One drawback of the AIS scale is that it 
essentially expresses the threat to survival, but not other aspects of the in- 
jury, such as degree or kind of resulting disability. 

A more detailed description of injury severity is the Occupant Injury 
Classification (OIC) . It is the best quantitative measure of injury severity 
available for evaluating FMVSS 214. It is available in a few existing data 



* 

NASS = National Accident Sampling System 
NCSS = National Crash Severity Study 



3-2 



bases (RSEP, NCSS) . The OIC is a five character code, one of which is the 
AIS. The other four characters represent body region, aspect, lesion, and 
system/ organ . The OIC would provide not only the most reliable measure for 
detecting shifts in injury severity, but it also would make it possible to 
distinguish between intrusion-related and non- intrusion-related injuries. 

The quantitative measure of FMVSS 214 performance is passenger compart- 
ment intrusion. The collision code used by most existing data bases is the 
Traffic Accident Data Project Scale (TAD) . It consists of an impact location 
code and a damage rating from 1 to 6. The TAD scale does not sufficiently de- 
fine the location of passenger compartment impacts for the purpose of evaluat- 
ing FMVSS 214. A more comprehensive collision scale is the Collision Deforma- 
tion Classification (CDC) which is available in the RSEP and NCSS data bases. 

The location of the impact is quite precisely defined by the CDC, but the ex- 
tent of deformation is not. The depth of intrusion is not directly defined by 
the CDC because of varying door widths and interior design. However, it may 
be derived by using the dimensions of the car. 

Means of Complying with the Standard 

FMVSS 214 was introduced in October 1970 with an effective date of Janu- 
ary 1, 1973. The manufacturers had been working on side door guard rails 
since at least 1968.* ** Various proposals were made as to the structural means 
of complying with the Standard, including the use of beams, structural foam, 
and honeycombed members. A review of present vehicle door constructions shows 
that the method of compliance is primarily the use of formed or channel-shaped 
metal beams of stampings positioned near or against the inner side of the out- 
er door sheet metal surface^, thereby providing the greatest resistance to in- 
trusion for the prescribed force application of FMVSS 214. Attachment of the 
reinforcing beams consists of spot or seam welds to the vertical door frame 
members on the hinge and latch sides of the doors. This method of reinforcing 
the doors is probably universal in the thin structured doors of small cars. 

Some of the larger vehicles, having a large door thickness between inner and 
outer panels, appear to accomplish the strength requirements by incorporating 
heavy metal frames within the door which are functional in supporting the win- 
dow regulators and latch mechanisms, thereby reducing the cost of additional 
structure for the sole purpose of increasing door strength. 

The Standard requires loading for 18 inches of crush. After about 6 inches 
of deformation, the reinforcement side beam has lost its ability to resist ad- 
ditional load as a beam. Its resistance to side crush becomes a function of 
the tensile strength of the beam concentrated at the end attachments. Thus, 
the strength of the door frame and hinge attachments become the critical design 
features for intrusion of more than about six inches. 

* 

RSEP = Restraint Systems Evaluation Project. 

Hedeen, C. E. and D. D. Campbell (Fisher Body Division, General Motors Corp.), 
i Side Impact Structures . Society of Automotive Engineers, 1969. 

The domestic manufacturers use channel beams with corrugated logitudinal rein- 
forcing and sometimes center plate reinforcement. Volkswagen has used a sim- 
ple channel beam on their newer models; however, in the VW Beetle the beam 
flanges narrow at the connection point, which may reduce their effectiveness 
in off-center or angle side door collisions. 



3-3 



Primary and Secondary Effects of Compliance 

Side door beams significantly reduce occupant compartment intrusion in low 
speed impacts. From physical analyses it appears that strengthened door con- 
struction has increased effectiveness of occupant protection in the case where 
vehicles strike a glancing blow into the center door span, due to the low velo- 
city normal to the door surface at a given impact speed and the likelihood of 
deflecting the striking vehicle at relatively low impact speeds (below 15 mph) . 
This could prevent vehicle entanglement and loss of driver control which might 
cause more serious secondary collisions. Primary factors in considering the 
overall protection afforded by improved side door strength are (1) the relative 
weights of the vehicles involved in a glancing collision; (2) the relative velo- 
city of the striking vehicles; (3) the angle of impact and the front corner con- 
figuration of the striking vehicle; and (4) the vertical location of the door 
reinforcement in the struck vehicle. 

The most important unintended secondary effect is that the stiffening of 
the side door increases the acceleration forces on occupants in light-weight 
vehicles struck at relatively low speeds. Other possible secondary effects 
are less certain. In sideswipes, the side door beam may deflect the striking 
vehicle rather than absorbing the kinetic energy and slowing the striking 
vehicle. In certain types of collisions, it is possible that the beam could 
come free and become an injury-producing object. Also, the addition of side 
beams should enhance the integrity of the compartment in higher speed frontal 
collisions . 

Real-World Performance of the Standard 

The major factor affecting the relation between FMVSS 214 and real-world 
crashes is the static nature of the impact test. This limits the representa- 
tiveness of the test to a narrowly defined set of crash configurations. There 
are many variables involved which influence occupant injury, but the assumption 
is that the test specifications delineate the critical ones. Thus, if the test 
specifications of the Standard are met, then a significant improvement in oc- 
cupant crash protection is provided. The evaluation methodology must test this 
assumption . 

FMVSS 214 requirements are based on assumed relation between depth of in- 
trusion and occupant injury. Injury may be caused by the vehicle door intrud- 
ing upon the occupant as well as by the occupant's striking the door and/or 
other parts of the car, or other occupants. Intrusion of the door is depen- 
dent on the force of the impact, as is the force with which the occupant hits 
elements of the vehicle interior. It is not directly obvious to what extent 
the observed correlation between intrusion and injury reflects a causal effect 
of intrusion rather than their both being a result of the common force of im- 
pact. Therefore, it is not sufficient to restrict the evaluation to studying 
the depth of intrusion. It is also necessary to study injury reduction with 
respect to all relevant pre-crash and crash factors. 



3-4 



Some of the relevant factors which might be considered are: vehicle load- 

ing, road conditions, duration and degree of braking and/or rolling, and energy 
absorbed in vehicle rotation after impact. Injuries may be related to vehicle 
seating arrangements, occupant distance from the door, the shape of the in- 
terior surfaces, and the number of passengers seated adjacent to one another. 
The obvious factors of vehicle weights, relative velocities, body types, and 
occupant age, size/weight, and restraint-use must be considered. 

3.2 Review of FMVSS 215 - Exterior Protection 



This Standard has changed considerably since it first become effective on 
September 1, 1972. The increasingly stringent crash test requirements created 
considerable difficulty and there were numerous modifications and exemptions, 
especially for specialty cars (sports, vintage, etc.), In March 1976 a new 
Bumper Standard (Part 581 of Title 49) was issued under the authority of Title 
I of the Motor Vehicle Information and Cost Savings Act. Manufacturers present- 
ly can comply under either FMVSS 215 or Part 581; however, beginning September 1, 
1978, Part 581 is mandatory, with its broader damageability standards. Table 
3-1 below shows the major changes to FMVSS 215 as they apply to vehicle model 
years . 



TABLE 3-1 

APPLICABILITY OF THE STANDARD BY MODEL YEAR* 



Model 

Year 


Exterior Protection Standard Requirements 




pre-1973 


• No requirements. 


1973 


t 5 mph front; 2.5 mph rear barrier crash. 


1974 


• Horizontal pendulum test added over 115" wheelbase. 


• 


• Rear barrier crash increased to 5 mph. 


1975 


• Number of horizontal pendulum impacts reduced to 2 




front and rear. 




• Horizontal pendulum test for all cars. 


1976 


• Corner impact test for cars less than 1 20 wheelbase. 


1977 


• Corner impact test for all cars more than 120" wheel- 




base. 


1979 


• FMVSS 215 superseded by Part 581 - Bumper Standard t 




which increases damageability standards. 



♦Some changes in the Standard may have gone into effect after the 
start of a model year so that in that year some models may not have 
satisfied the Standard. 



3-5 



Purpose of FMVSS 215 



• The specific purpose is to establish requirements for impact re- 

sistance and the configuration of front and rear bumpers. 

• The general purpose is to prevent low-speed accidents from impair- 

ing safe operation of the vehicle and to reduce the frequency of 

override and underride in higher speed collisions. 

[The new Bumper Standard (Part 581) deals with reducing all 
physical damage to the front and rear of the vehicle.] 

General Requirements of FMVSS 215 

The current Standard requires both pendulum and barrier crash tests. 

Earlier versions (see Table 3-1) exempted certain vehicles or had lower criteria. 
Generally, the test conditions are: 

• Two pendulum tests 

The longitudinal impact test consists of impacting the front 
and rear bumper surface two times each at 5 mph with an im- 
pacting mass equal to the weight of the vehicle. 

The corner impact test consists of impacting the front and 
rear corner twice each at 3 mph at an angle of 60 degrees 
from the longitudinal centerline of the vehicle. 

a Barrier test 

Two fixed barrier collisions with the vehicle traveling at 5 
mph, once forward, once in reverse. 

Generally, the protective criteria are that safety equipment not be impair- 
ed; hood, trunk and doors operate normally; there are no leaks from fuel, cool- 
ing, exhaust or energy-absorbing systems; vehicle mechanical systems remain nor- 
mal; and that the test device impact only on its impact ridge. 

Measures of Effectiveness 

The primary purpose of the bumper Standard FMVSS 215/Part 581 is to prevent 
low speed collisions from impairing the safe operation of vehicle systems and 
to reduce the frequency of override or underride in higher speed collisions. As 
a consequence, the cost of repairs to vehicles as a result of low speed collis- 
ions is expected to be reduced and economic advantages to the consumer would be 
realized directly through less cost and inconvenience of necessary repairs, and 
indirectly through reduced cost of insurance. Reduced damage in highway accidents 
could reduce traffic tie-ups and, hence, result in fewer secondary accidents. 

Performance measures used to insure that safety-related items are not ren- 
dered inoperable include pendulum and barrier impact testing of the bumper sys- 
tem. The safety-related requirements are: 



3-6 



• Reflectors not be cracked, and lamps (excepting license plate 

lights) not be damaged beyond adjustability. 

• Hood, trunk and doors operate in a normal manner. 

• Fuel and cooling systems develop no leaks or constrictions and 

caps and seals remain unaffected. 

• Exhaust systems develop no leaks or constrictions. 

• The propulsion, suspension, steering and braking operate in a 

normal manner. 

• The impact device should not strike the vehicle except along a 

specified impact ridge. 

• The energy-absorbing impact device should not suffer any loss of 

gas or liquid. 

Means of Complying with the Standard 

FMVSS 215 for front and rear bumpers has undergone considerable revision 
since it first became effective on September 1, 1972. The elimination or re- 
duction of damage resulting from low-speed impacts requires the application of 
the basic principle of energy absorption. A variety of approaches and method- 
ologies has been suggested and/or utilized including various torsional systems, 
mechanical systems, or energy-absorbing materials. The energy-absorbing materials 
used are springs, pneumatic shock absorbers, plastic foams, etc. 

A listing of the major means for compliance that have been used or suggest- 
ed include the following [1, 2, 3, 4]. 

• Full-width steel reinforcement behind a bumper attached to rubber 

block which is energy-absorbing. (Chrysler) 

• Steel beams on both sides of vehicle support steel bumper and are 

connected to energy-absorbing devices consisting of prestressed 

rubber (slabs which stretch or shear upon impact) . (Ford) 

• U-shaped steel bumper which contains energy-absorbing cellular 

plastic blocks in the interior of the bumper. (Saab) 

• Reinforced steel bumpers with external rubber guards attached to 

energy-absorbing hydraulic/pneumatic cylinders on either side of 

the car. (General Motors) 

• Soft-faced front end of elastomeric material such as urethane which 

is energy-absorbent. (General Motors) 

• Steel cable bumper decelerator which rides freely over car frame ex- 

tentions and alters the direction of energy absorption from longi- 
tudinal to transverse. 

Systems designed to meet the Standard can be classified as either (a) re- 
turnable: spring, spring and shock absorber (hydraulic), state-of-the-art 

bumper material (metallurgy) with or without any combination of the above, elasto- 
meric bumper materials with or without the above, or (b) non-returnable: shock 



3-7 



absorber types which are either rechargeable or reset by hand, or deformable 
energy absorbers which must be replaced after collision to bring them to 
their original manufactured state. The most frequently used compliance method 
in recent model years has been the returnable energy-absorbing hydraulic/pneu- 
matic cylinder. 

Primary and Secondary Effects of Compliance 

The primary effect of the Standard is to reduce or eliminate vehicle dam- 
age and prevent impairment to the safe operation of the vehicle for the follow- 
ing low speed (5 mph or less) crash situations. 

@ Front end, rear end and front and rear angular collisions with fix- 
ed objects at least the height of the bumper. 

& Head-on collisions between vehicles with equal bumper heights on 
a surface allowing them to be level with respect to each other 
(except for very large differences in mass of two vehicles) . 

• Collisions where bumper mismatch does not result when the rear col- 

liding vehicle is pitched due to braking, crown of road, and/or 
inclining or declining grade. 

9 Angular collisions between vehicles ( front- to-front , rear-to-rear 
and front-to-rear) that are level with respect to each other, 
within a maximum angle. 

A number of potentially significant secondary effects can be noted. The 
new bumper designs have more complicated interfaces with other systems such as 
the radiator, grille and lights. In higher speed crash situations not covered 
by the Standard, the cost of damage sustained to the bumper and interface com- 
ponents may be higher. Because of the greater protrusion of some new bumpers 
which meet the Standard, the complying vehicle may cause greater damage in 
higher speed collisions. 

Real-World Performance of the Standard 

Comparison of the desired effects of Standard FMVSS 215 indicate the fol- 
lowing areas to be considered in actual vehicle operating conditions. 

• The desired bumper match may not occur under the conditions of un- 

even roadways; particularly on crowned roads at intersections, and 
also when there is considerable vehicle pitch due to weight trans- 
fer caused by acceleration and braking. Also, a dangerous load 
mismatch may occur when a bumper end strikes another bumper sur- 
face at an angle causing high unit load force and local deformation. 

• The strengthened bumper may cause more severe penetration into the 

side and door structure of other vehicles at both low and high 
speed side impacts. 



3-8 



• Five mile per hour impact damage may result in extensive vehicle 

structural damage depending on bumper configuration and attach- 
ment methods employed, even though safety-related items are un- 
damaged. This moat probably might occur on unibody type vehicles 
having reduced strength capability at the bumper bracket attach- 
ment locations, as in smaller cars with relatively light frames. 

• With the wrap-around projecting bumpers, "hooking" a front and 

rear bumper becomes a hazard. 

3.3 Review of FMVSS 301 - Fuel System Integrity 

Since its introduction in 1968, this Standard has been modified several 
times, increasing the difficulty of meeting the test criteria. For example, 
the static rollover test was first proposed in 1973 for the 1976 models; that 
test requirement was temporarily suspended, while new test criteria were con- 
sidered. The 1976 models had to meet the frontal crash and static rollover 
requirements. The present 1977 models must meet front, side, and rear barrier 
crashes as well as static rollovers. Vehicles in the 6,000 and 10,000 pound 
GVWR* (typically multipurpose vehicles such as vans or pickups) must meet the 
passenger car requirements by the 1978 model year. Table 3-2 describes the 
applicability of the Standard by model year. 

Purpose of FMVSS 301 

• The specific purpose is to establish requirements for the integrity 

of motor vehicle systems. 

• The general purpose is to reduce deaths and injuries occurring from 

fires resulting from fuel spillage in motor vehicle accidents [5]. 

General Requirements of FMVSS 301 

• In the barrier tests for fuel spillage, the vehicle must not lose 

more than: 

- One ounce by weight during the crash. 

Five ounces during the next five minutes after the crash. 

- One ounce in any one minute period during the next twenty- 

five minutes. 

• In the rollover test, fuel spillage is limited to five ounces in 

the first five minutes at any 90° increment or more, and is limit- 
ed to no more than one ounce during any subsequent one minute period 
while the vehicle is at rest. 

• Currently, passenger cars (1977 model) must undergo 30 mph front 

barrier and rear moving barrier crashes, a 20 mph lateral moving 
barrier crash and a static rollover. 



Gross Vehicle Weight Range. 



3-9 



TABLE 3-2 

APPLICABILITY OF THE STANDARD BY MODEL YEAR 



Model 

Year 




★ 

Fuel System Integrity Requirements Set by FMVSS 301 


Pre-1968 


« 


No requirements 


1968 


• 


Frontal barrier crash (30 mph) and limited leakage from 
fuel tank, filler pipes, and fuel tank connections dur- 
ing impact (one ounce) and after impact (one ounce per 
minute). Effective January 1, 1968. 


1971 


• 


In response to air pollution control legislation, auto 
manufacturers installed evaporative emission-control 
systems increasing fuel system elements. 


1976 


• 


Passenger cars must meet front barrier impact and static 
rollover test. 


1977 


• 


Side and rear barrier impact tests are added to passenger 
car requirements. 




• 


Other vehicles up to 6,000 pounds GVWR must meet 1976 
passenger car conditions plus the rear impact test. 




9 


6,000 to 10,000 pound GVWR vehicles must meet only the 
front barrier test. 


1978 


• 


All vehicles up to 10,000 pounds GVWR must meet the 1977 
passenger car requirements. 



The 1976 modifications were announced in 1973 and manufacturers had considerable 
lead time to introduce improvements in pre-1976 models in anticipation of the 
effective date of the Standard. 



• The 1977 model year multipurpose vehicles of less than 6,000 lb 

GVWR must undergo only the perpendicular front barrier crash, the 
rear moving barrier crash, and the static rollover. The 1978 
models must meet the current passenger car criteria. 

® The 1977 multipurpose vehicles of between 6,000 and 10,000 lb GVWR 
must meet the perpendicular front barrier crash criteria. The 1978 
models must meet the current passenger car criteria. 

e School buses, which are 10,000 lb GVWR or greater, have to meet a 

special moving contoured-barrier crash test starting July 15, 1976. 
The evaluation of the effectiveness of this Standard with regard 
to these school buses is not within the scope of this project. 

The static rollover test occurs after an impact test. The vehicle is rota- 
ted about its longitudinal axis in 90° increments. Each incremental rotation 
should take between one and three minutes and the vehicle should remain in each 
position for five minutes. 



3-10 



Measures of Effectiveness 



There seems to be no direct, quantitative scalar measure which relates 
accident conditions to the effectiveness of this Standard. Using the Abbre- 
viated Injury Scale (AIS ) , police or accident investigators would have to 
classify burns and asphyxiations separately from other injuries. For instance, 
AID-1 includes all first degree burns or some second degree burns. It also 
applies to minor aches and sprains. An occupant may suffer slight (AIS-1) 
bruns and more severe (AIS-2) bodily injuries. However, normally only one 
injury (the most serious) classification is designated for each victim in a 
crash. This would decrease the effectiveness of using existing AIS data with 
regard to burns. Use of vehicle deformation or any other such impact measure 
(vehicle speed, direction and location) adds the factor of "indirect" collis- 
ions — that is, the initial impact causes some other part of the vehicle to 
impact and damage the fuel system. 

The most promising approach to evaluating FMVSS 301 may be to combine 
various effectiveness measures such as: fire-caused deaths in auto collisions 

as a percent of all fatal accidents, or the rate of fuel system ruptures in 
the towaway accident population. Neither measure alone is likely to directly 
reflect the effect of the Standard. Deaths due to fire in auto accidents may 
increase (or decrease) because of better (or worse) escape conditions, mater- 
ials giving off toxic fumes, etc. Ruptured fuel systems in towaway accidents 
may represent a biased sample of accidents and the number of fires may increase 
or decrease, depending on the ignition sources. Also, there is the further 
possibility that the fire (and subsequent injury or death) may not be due to 
the occupant’s vehicle but to some other vehicle. For example, cars striking 
exposed fuel tanks on trucks may result in fire and injury in the striking 
vehicle. 

Means of Complying with the Standard 

A variety of approaches, most of which can be implemented in concert, 
have been suggested for compliance. The means of compliance are briefly listed 
below and are discussed in References 6, 7, and 8. 

• Fuel Tank Location: For a front-engine vehicle the most protective 

location would be the area between the rear wheels above the rear 
axle and below the rear window. The regions close to the rear 
fender or either side of the car are more vulnerable to rear end 
or side impacts. (Mercedes and the VW Dasher have protected or 
interior fuel tanks, as do many U. S. station wagons.) 



*The plastic materials being used to lighten new cars increase the available com- 
bustible material and burn at an intense heat, thus increasing the hazard to 
occupants, once a fire is initiated. 



T 



• Fuel Tank Material and Shape : Horizontally aligned rectangular 

flat tank configurations with smoothed contours and corners of- 
fer the least hazardous design. The strength of tank walls 
should take into account fuel capacity and size of car. Alter- 
natives to rigid metal construction include plastic fuel tanks 
and expandable tanks with corrugated folds which permit altering 
the geometric shape of the tank [6] . 

9 Fuel Tank Anchorage : The straps and anchor points for the tank 

must be sufficiently strong to withstand extreme distortion and 
inertial forces associated with impact. 

9 Filler System : In general, the protrusion of the filler neck from 

the tank should be as short as possible, consistent with the loca- 
tion of the tank. The major change that manufacturers made to 
initially satisfy the Standard was to upgrade the filler tank cap. 
Self-sealing breakaway type fittings have been suggested for the 
filler system and the other outlets from the fuel tank. The vapor 
vents have float valves to prevent fuel leakage but these could be 
defeated in rollover accidents. 

o Vent Line and Fuel Line : As mentioned above, it has been suggested 
that all fittings to the fuel tank be of a self-sealing breakaway 
type. In addition, the location, length, flexibility and strength 
of the vent and fuel lines all affect the possibility of rupture 
and fuel leakage. 

9 Carburetor/Fuel Pump/Fuel Filter Locations : The location of these 

components in the front end relative to other systems will influ- 
ence successful compliance with front or lateral moving side bar- 
rier tests. 

Primary and Secondary Effects of Compliance 

"Even a cursory review of contemporary designs shows that fuel systems 
have not been considered as a single, integrated, rupture-resistant system, but 
as a set of components adapted to a particular vehicle after its basic design 
has been completed" [9]. The major effects of the Standard have been the re- 
positioning of the fuel tanks and filler spouts and the upgrading of the fuel 
filler cap. The repositioning of the tank might have some secondary effect on 
the performance of motor vehicles, because it changes the weight distribution. 
However, this would be hardly perceptible and probably beneficial. Reposition- 
ing the fuel tank to more interior parts of the car would increase the hazards 
to the occupants in the case of a fire (though the probability of fire and leak- 
age may be reduced). Thus, most design change recommendations include fuel 
tank repositioning and introduction of a fire wall for protection of rear seat 
passengers . 



3-12 






Another secondary effect, at least partially ascribable to the Standard, 
is the increased complexity of the carburetor.* The system has become more 
enclosed and more difficult to service, partly to prevent leakage from the 
carburetor during the rollover test. 

For Multipurpose Vehicles (MPVs) , there has been rapid design develop- 
ment to meet the Standard. With the greater weight, longer fuel lines, and 
lack of energy absorbing bumper systems of MPVs it is more difficult to control 
fuel leakage in frontal crash tests. To meet the Standard, MVPs may require 
structural changes which passenger cars do not need. 

Real World Performance of the Standard 

It is clear that the specifications of FMVSS 301 do not directly apply to 
a number of crash situations. These include; 

• Those at speeds higher than specified in the Standard. 

6 Impacts with any object which is not perfectly flat (poles, abut- 
ments, car bumpers, etc.). 

• Real world rollover crashes, especially where the filler spout pro- 

jects out from the vehicle body. 

• Collisions causing intrusion into the area of the fuel tank, filler 

spout or evaporative canister. 

• Running off the roadway over barriers or rocky, uneven terrain. 

In general, fire and/or fuel spillage are relatively rare events in motor 
vehicle collisions [9, 10, 11]. The various studies summarized in Reference 9 
point out an important fact in evaluating the real world performance of this 
Standard: fire occurs in approximately one in a thousand motor vehicle acci- 

dents, and only one in twenty of all vehicle fires is due to a collision. 

Given these figures, there are about 17,000 accident-related vehicle fires per 
year in the entire country; and of the vehicle fire records which fire depart- 
ments might keep, only 5 percent of their reports would apply to vehicle fires 
due to collision. The measurement of the more frequent occurrence of fuel 
spillage is harder to detect because of evaporation and absorption of the lost 
fuel. The frequency of fuel system damage in real world accidents is perhaps 
the best physical measure of an indirect effect of the Standard. 

Because there is an obvious relationship between fires, fuel sources and 
ignition sources, the real world performance of the Standard will depend on 
limiting potential interactions between the fuel and ignition sources. There- 
fore, the impacts of the introduction of the fuel vapor recovery system and 
catalytic converter, as well as a consumer trend toward purchase of vans, motor 
homes and other potentially hazardous larger vehicles, makes the evaluation of 
the performance of the Standard even more difficult. 



The majority of the changes to the carburetor have resulted in engine per- 
formance improvement. 



3-13 






3.4 Review of FMVSS 208 - Occupant Crash Protection 



Originally introduced in 1968, the Occupant Crash Protection Standard 
has been modified several times. Its major change has been to allow vehicle 
manufacturers three options for satisfying the Standard. Options //I and //2 
have less specific equipment criteria and more detailed injury criteria. Op- 
tion // 3 has specific equipment requirements for the seat belt assemblies but 
few or no injury criteria, depending on the type of assembly installed. The 
objective of this Standard is to decrease occupant injury through increased 
usage of restraint systems — active systems such as the current lap/shoulder 
belt combination, or passive system typified by the passive belt or air cush- 
ion restraint system. In many of the earlier versions of the Standard, the 
active methods of occupant crash protection were scheduled for elimination. 
There has been considerable controversy concerning the relative effectiveness 
and costs of the alternative active and passive systems. The current version 
of the Standard does not give any date for the elimination of active systems. 
Since the Standard became effective on 1 January 1968, automobiles have been 
equipped with a variety of occupant restraint systems, such as lap belt only, 
separate lap belt and shoulder belt, and integral lap belt and shoulder belt. 
At present, the overwhelming majority of vehicles have the integral lap belt 
and shoulder belt system. Table 3-3 gives the important changes in the Stan- 
dard by model year. 

Purpose of FMVSS 208 

• The specific purpose is to establish performance requirements 

for the protection of vehicle occupants in crash situations. 

• The general purpose is to reduce the number of deaths and the 

overall severity of injuries in motor vehicle accidents. 

General Requirements of FMVSS 208 

The current Standard allows the manufacturer to comply under three dif- 
ferent options, each with different performance criteria. In general, the 
requirements are: 

• Option //I requires a completely passive protection system which 

meets all the injury criteria in the frontal barrier crash at 
30 mph and the lateral moving barrier crash at 20 mph. In the 
rollover test at 30 mph the only injury criterion is that the 
test dummy should be contained within the passenger compartment 
throughout the test. Other injury criteria limit the forces on 
the head, chest and upper leg during crash tests. 



The effectiveness of the Standard depends completely on the usage of the pro- 
tection systems. The passive system is favored because it would always be in 
use, without an explicit action ("buckling up") on the part of the occupant. 



3-14 



TABLE 3-3 

APPLICABILITY OF THE STANDARD BY MODEL YEAR 









Model 

Year 


Occupant Crash Protection Standard Requirements 


[ 


Pre-1968 


• No requirements, but lap belts were standard equipment on 
most cars. 




1968* 


i Type 1 (lap) or Type 2 (lap and shoulder) seat belt assemblies 
required at each seat position. (FMVSS 209 specifically de- 
scribed the assembly and FMVSS 210 described requirements for 
the anchorage.) 


. 


1972** 


• Manufacturers were given three options for meeting the Standard. 
The first option required a totally passive system for crash 
protection. The second option required a lap belt and some 
other passive features to meet the frontal crash requirements. 
The third option specified an Integral lap/shoulder belt 
system with warning device and had no Injury criteria. (After 
August 15, 1973, the third option was to be eliminated! however, 
that date was continually postponed. 




1974 


t The third option was modified to require an Ignition Interlock 
device. 

t If only a lap belt is used, the vehicle had to meet the frontal 
barrier crash requirements and Injury criteria. 

0 The second option was upgraded to a complete passive protection 
system In head-on test crashes although some type of seat belt 
was still required. 




(1975) 


(0 The Ignition Interlock requirement was revoked early In the 
1975 model year— 29 October 1974. However, many models were 
produced with the interlock system.) 





*FMVSS 208 became effective 1 January 1968, which was after the beginning 
of the 1968 model year. 

This change came after the start of the 1972 model year (1 January 1972); 
however, this change did not affect how the manufacturers were complying. 



• Option #2 requires a head-on passive protection system for front 

seating positions which meets all the injury criteria in a 30 
mph perpendicular, frontal barrier crash. The option also re- 
quires installation of at least a lap belt with warning system. 

• Option #3 requires only a lap and shoulder belt protection system 

with a belt warning system. If only a lap belt is provided, then 
the vehicle must be capable of meeting the perpendicular frontal 
barrier crash requirements including injury criteria. 



3-15 



Measures of Effectiveness 



Since the Standard's stated purpose is to reduce the occurrence and sever- 
ity of injury, injury-related measures are the most obvious means of assessing 
the Standard's effectiveness. The injury criteria employed for testing under 
the Standard are: 

• The test dummies used in each crash test are to be contained with- 
in the passenger compartment throughout the test. 

c The acceleration of the head of the test dummies cannot exceed an 
index level of 1,000. The index is an integrated expression of 
the acceleration forces on the head in any period during the crash. 
Prior to 31 August 1976, the acceleration was measured during any 
period when the head is in contact with any part of the vehicle 
other than the belt system. 

0 The acceleration forces on the chest are measured at the center of 
gravity of the upper thorax. These forces must not exceed 60g 
for longer than 3 milliseconds total. Prior to 31 August 1976, 
this acceleration was measured with a severity index which could 
not exceed 1,000. 

® The axial forces on the upper leg cannot exceed 1,700 pounds. 

The above explicit injury criteria, however, are applicable only under 
the first two options for passive protection systems.* The vast majority of 
automobiles in recent model years (1973-1977) are equipped with seat belt 
assemblies which comply with the third option and thus the net effectiveness 
of this restraint system depends on their usage by vehicle occupants. For this 
reason, the estimating of the effectiveness of the Standard must cover both the 
effectiveness and usage of the system. Because the Standard's stated purpose 
is the reduction of the number and severity of injury, the Abbreviated Injury 
Scale (AIS) is the most obvious measure of effectiveness of the Standard. 

Means of Complying with the Standard 

Since 1 January 1972, manufacturers have had three options under which 
they could comply with FMVSS 208. The first option was to provide a totally 
passive system: no manufacturer has complied under this option. The second 

option encourages the manufacturer to provide some passive protection systems, 
but does not require complete reliance on the passive systems as the first op- 
tion does. Option #2 requires, when using the passive system alone, that in- 
jury criteria must be met for front seat passengers in frontal collision into 
a barrier at 30 mph. However, these vehicles are also required to have seat 
belt assemblies with warning systems, with some exceptions in the case of pass- 
ive belts. Some manufacturers have provided systems which have met this op- 
tion on some of their cars. General Motors provided an Air-Cushion Restraint 
System (ACRS) as an option on a few of their larger vehicles for several model 



With the exception that under Option #3 , if only a lap belt is provided, then 
the vehicle must be capable of meeting the perpendicular frontal barrier crash 
requirements, including injury criteria. 



3-16 



years. Volvo is currently field testing an air bag type system on some of 
their cars. Since 1975, Volkswagen has offered a passive belt system as an 
option in its VW Rabbit. 

The vast majority of cars sold in the U. S. today comply with FMVSS 208 
under the third option — combination lap/shoulder belt assemblies with warning 
devices. If a manufacturer chooses to provide just a lap belt, then he has to 
show that the vehicle meets the perpendicular frontal crash test requirements, 
which include injury criteria. By providing the lap/shoulder belt combination, 
the manufacturer has only to meet hardware requirements, not crash performance 
criteria. The seat belt assemblies must fit a wide range of persons. The lap 
belt portion must fit everyone from a 50th-percentile 6-year old to a 95th-per- 
centile male (i.e., 47 to 215 lbs, respectively). The shoulder portion must 
fit everyone from a 5th-percentile female to the 95th-percentile male with 
the seat in any position. The lap belt portion must have an emergency-locking 
or automatic-locking retractor, while the shoulder portion must be adjustable 
manually or with an emergency-locking retractor. 

The seat belt warning system has many detailed specifications about when 
and how it should operate. During the 1974 model year and part of 1975, the 
seat belt warning/ignition interlock system stirred considerable controversy. 

The interlock requirement was revoked by Congress in 1974. Presently, both 
a visible and an audible warning are given for at least four and not more than 
eight seconds when a seat is occupied and the belt is not buckled. 

Since introduction of the Standard, there have been several variations of 
the seat belt restraint system in cars sold in the U. S. Table 3-4 below des- 
cribes by model year the method used in most models. 

Real-World Performance of the Standard 

The real world performance of FMVSS 208 is dependent on a number of key 
factors which can be grouped under the following headings: (1) Usage; (2) Char- 

acteristics of Occupants; (3) Actions of Occupants; (4) Characteristics of Car 
Interior; and (5) Type of Accident. 

Usage . The overwhelming majority of cars complies with FMVSS 208 through 
the inclusion of active restraint systems which require action on the part of 
the driver and other occupants. A significant majority of drivers and passen- 
gers does not use the system, and, hence, considerably negates any potential 
benefits in terms of injury reduction or elimination which could accrue from 
the Standard. Urban usage surveys suggest that usage is 20 to 30 percent. 



3-17 



TABLE 3-4 

PRIMARY CRASH PROTECTION COMPLIANCE METHODS 



Model Year(s) 


Common Type of Seat Belt Assembly 


1968 - 1971 


e Domestic manufacturers supplied cars equipped 
with lap belt systems. Some provided 
additional shoulder belts. 

[Foreign manufacturers often supplied a 
Type 2 (3-point) belt.] 


1972 


® Late model year cars came equipped with a 
persistent belt warning system. More 
domestic manufacturers supplied separate 
lap belts (Type 1) and shoulder belts 
(Type 2a)-- a 4-point system. 


1973 


• The Standard required a Type 2 belt with a 

detachable shoulder portion. 

. . 


1974 - 1975 


• Ignition interlock was introduced to be used 
with Type 2 belts (non-detachable shoulder 
belts). The persistent warning system was 
changed to a simple (4-8 second) warning 
system in early 1975 model year cars. 


1976-Present 


e Although the ignition interlock requirement 
was revoked early in the 1975 model year, 
the interlock system was not removed from 
most cars until the following model year. 

1 



Characteristics of Occupants . Requirements for the seat belt assembly 
are that (1) the lap portion must fit persons from a 50th-percentile 6-year 
old to a 95th-percentile male (47 lb to 215 lb) and (2) the upper torso re- 
straint must fit all persons between a 5th-percentile female and a 95th-per- 
centile male with the seat in any adjusted position. Persons outside these 
ranges may find it difficult to make use of the restraint system and/or could 
experience seat belt-related injuries, if used. Even with properly adjusted 
belts, the flexing of the flesh and the type of clothing worn affect belt 
restraint effectiveness. 

The potential for occupant injury is, of course, affected by other oc- 
cupant characteristics. Occupant health, age and sex may have a significant 
effect. The very old and the very young can experience more severe injuries 
than a healthy adult in his or her middle years, for example. Tall people 
have an increased potential for head injury, especially in small cars. 



3-18 



Actions of Occupants . A number of actions taken prior to and during an 
accident can affect injury risk with the use of lap and/or shoulder belts. 
Loosely worn and improperly adjusted belts negate the load-limiting effects 
of belts and may cause additional injuries due to the belt. The retractable 
3-point lap/shoulder belt system reduces the likelihood of an improperly 
worn belt in the front outboard seating positions. 

Proper seating position will affect the potential for the restraint sys- 
tem to protect an occupant from injury. Obviously, when an occupant is lean- 
ing forward or sitting sideways, the lap/shoulder belt system may be ineffec- 
tive or less effective in preventing injury. 



Characteristics of Car Interior . The effectiveness of belt restraint 
in minimizing injuries will be affected by the quality of instrument panel 
padding and bending and/or fracture strength. This is covered by FMVSS 201. 
The adjusted front seat position regulating the distance from the driver/ 
passenger to the steering wheel/front dashboard is another factor affecting 
possible injuries. Other factors such as an open glove compartment or ash 
tray or loose objects can contribute to injuries. 



Type of Accident . The action and potential effectiveness of restraint 
systems in reducing or preventing injury are related both to type of injury 
and collision speed. At very low speeds, there is usually no injury, while at 
extremely high speeds, all occupants are usually killed or injured, often because 
of destruction or major deformation of the passenger compartment, occupant 
ejection, or fire. Seat belts are expected to have their greatest effectiveness 
at moderate speeds. 

The type of impact is also important. Rear collisions cause rearward 
neck strain which is not addressed in the Standard. In this case, the back 
of the seat and head restraint comprise the restraint system. The effective- 
ness of belt restraint in frontal and side impacts may be quite different, due 
to significant differences in the lateral and longitudinal loading forces. 

3.5 References for Section 3.0 



1. Larousse, Rene, Energy Absorption by Structural Deformation, SAE Report 

No. 73007, January 8-12, 1973. 

2. Hai, Wu, A Study of Automotive Energy Absorbing Bumpers , SAE Report No. 

730024, January 8-12, 1973. 

3. Compton, R. H. , C. Westphal, Jr., and R. Crone, Bumper Systems Soft Face 

vs. Model 1973 Steel System, Motor Vehicle Programs (NHTSA) , November 
1974. 



3-19 



Hi 



4, Appleby, Michael R. , Occupant Safety and Damageability Considerations 

Related t& 1074 Automotive Bumpers 3 SAE Report No. 740989, October 

21-23, 1974, 

5, NHT8A, Fart 571 Motor Vehicle Safety Standards: Standard No. 301 - Fuel 

System Integrity , Pfderal Register , Vol, 40, No. 200, October 15, 1975, 
p, 48353, 

6, Sevary, D»> 0, Blaisdell, and J. Kerkhoff, Automobile Collision Fires 3 

Proceedings pf the Eighteenth STAPP Cay Crash Conference . Society of 
Automotive Engineers, Inc., 1974, 

7, SlQbartaon, 8, h,, A New Look at Fuel System Design Criteria 3 Proceedings 

pf the Tepjfy §TAPP Car Crash Conference . Society of Automotive Engin- 
eer e, Inc, ,1061, 

8, Savory, p, M» , H, M. Brink, and J. D. Baire, Postorash Studies Show Need 

for Rear-Seat Fir® Wall and Rupture-Proof Fuel Tank 3 SAE Journal , 

July 1968, 

9, Cooley, P,, Fire in Motor Vehicle Accidents 3 HIT LAB Reports . Highway 

Safety Research Institute, Volume 5, Number 1, September 1974. 

10, Siegel, A,, and A, Nahum, Vehicle Postcollision Considerations 3 1970 

International Automobile Safety Conference Compendium . Society of 
Automotive Engineers, Inc., 1970. 

11. Johnson, N., An Assessment of Automotive Fuel System Fire Hazards 3 DOT- 

HS-8Q0-624, Dynamic Science, Phoenix, 1971, 



3-20 






4.0 APPROACHES TO EVALUATING THE STANDARDS 

SUMMARY 

The approaches to evaluating the Standards all face a similar problem: 
isolating the effect of one Standard from the effects of other Standards , 
changes in the Standard of interest , other changes in vehicle design and ma- 
terial , and external factors influencing accidents and severity . These prob- 
lems are approached by selection of existing data bases or collection of new 
data which promises to show the expected effect most clearly . All the sugges- 
ted approaches for evaluating the effectiveness of individual Standards had 
separate analyses of existing data and of new data . The mayor approach and 
problem for evaluating each Standard are: 

FMVSS 214 : Use detailed NCSS accident data and possibly gather simitar new 

data and use a statistical model to estimate the effect of the side door 
beam on injury and intrusion . Because of the complex nature of the in- 
jury mechanism and the engineering evaluation that the beam only has an 
effect at low speeds (and possibly a counteracting effect in some situ- 
ations) , the effect of the Standard may be difficult to isolate. 

FMVSS 215 : Using the State Farm Mutual Insurance Company claim data will show 

an initial estimate of the effect of improved bumpers on the frequency of 
damage to related parts. Analyses of other existing data bases cannot 
provide as clear a picture of the Standard’s effect because its effect is 
in tow speed normally non-reported accidents „ To delineate those acci- 
dents, we suggest a car owner survey. 

FMVSS 301 : Analysis of this Standard is hampered by the low frequency and re- 

porting inconsistencies of fire/fuet spillage accidents. Me feel that 
the most promising approach would be to check fuel system rupture in tow- 
aiway accidents . However, alt the approaches to evaluating this Standard 
— analyses of fire /police department data, and of fire-related fatalities , 
as welt as fuel system rupture — are speculative. 

FMVSS 208 : The basic approach for evaluating lap and lap/shoulder belts is to 

extend the RSEP study by combining that data base with the NCSS data. 
Secondly , BEV is being added to the RSEP data so that this new analysis 
can study the effect of impact speed. The analysis of the passive restraint 
system uses the same statistical model but must wait until sufficient data 
become available. 

Because many of the approaches use similar data and because of other rea- 
sons, it is possible that the evaluation of the Standards could be integrated 
to some extent, the most obvious cases of this being the uses of mass accident 
data, NCSS/RSEP data, towaway accident data, and hardware cost data. In the 
case of hardware costs, we have expressed some reservations that actual con- 
sumer out-of-pocket costs for a Standard are strictly related to hardware costs 
because of manufacturing and marketing price policies. 



4-1 



4.1 Approaches for Evaluating Individual Standards 



FMVSS 214 

The requirement for strengthened side doors is based on the experience 
that injury severity increases with depth of door intrusion in side impact 
crashes. Therefore, the performance requirement of the Standard is to limit 
the door intrusion in a crash. The ultimate purpose, however, is to reduce 
injury severity. If the Standard is successful, injury frequency will also 
be reduced, because minor injuries will be reduced to no injury. 

The injury generating mechanism is complicated. If a car is hit by 
another car, the door is deformed until the reaction forces are strong enough 
to move the car. Calculations suggest that initially the door structure is 
moving toward the occupant. Later, when the vehicle is moving sideways, the 
occupant moves relative to the vehicle and will finally hit the vehicle struc- 
ture somewhere, and possibly eject. The situation is similar when a car skids 
into a fixed object sideways. Since the side beam affects only one aspect of 
the injury mechanism, its effect may not be very obvious. Also, it may be 
limited to only certain types of injuries. 

The objective of the evaluation of the effectiveness of the Standard is 
two-fold: (1) to evaluate the performance reduction in intrusion, and (2) to 

evaluate the reduction in injuries. In both cases, it is clear that many 
factors other than side door strength influence the depth of intrusion and 
the forces on the occupant, and thereby the resulting injury. The most im- 
portant other factors are probably the speeds of the colliding vehicles, the 
angle between the directions of vehicle movement at the time of impact, and 
the exact point of door contact. Other factors are details of the construction 
of the vehicles, and the characteristics of the occupants such as height and 
weight. To make a valid comparison between cars with and without side beams, 
the effects of such factors have to be controlled in the analysis, or otherwise 
eliminated . 

The effects of the extraneous factors influencing intrusion and injuries 
are not sufficiently well known to eliminate them by analytical methods. 
Therefore, statistical methods have to be applied to empirically determine 
the influence of these factors and to eliminate them. There are several dif- 
ficulties in applying existing statistical techniques. One is that most of 
the factors influencing intrusion and injury are continuous, but some are 
categorical. However, in practice, some continuous variables are given only 
by categories. The combined use of categorical and continuous variables in a 
model poses a number of operational problems. A more serious problem in 
studying injury reduction is that injury is a categorical variable. Statis- 
tical analysis techniques which deal with categorical dependent variables can 
detect shifts from one category to another, but they cannot discern small but 
consistent shifts within several categories. An analysis limited to only two 
categories (e.g., "injury" and "no injury") may not be sensitive enough to 
detect small shifts over a wide range. 

If exactly one type or level of injury would result from any given com- 
bination of precrash factors, it would be relatively easy to determine the 
influence of these factors. In reality, however, the type and severity of 
injury resulting from a specific crash is not precisely predictable. The best 
one can expect is to predict the probabilities with which the various levels 



4-2 



or types of injuries occur. If the categories of "no injuries" and "injuries 
of low severity" are not completely reported, the estimates of these probabili- 
ties can be seriously distorted, and it might become impossible to detect a real 
effect of a Standard. The practical question is: how complete are "no injury" 
and "low injury" crashes reported? The success of any analysis that uses "fre- 
quency of a certain injury level" hinges on the answer. 

One way to overcome this problem is to restrict the analysis to towaway 
crashes. Need for towaway appears to be a fairly objective criterion for the 
severity of damage to a car. There exists, however, the possibility that side 
beams might reduce intrusion, and thereby reduce the need to tow a car, even 
though side beams may not reduce injury severity. In this case, reduction of 
the number of towaway crashes, and no change in injury severity in cars which 
are towed, may result in an apparent spurious increase in injury severity in 
side beam cars. 

Other approaches which can be considered are: 

• Studying risk of occupant injury per exposure measure. However, 

vehicle-miles-of-travel can only be measured with low accuracy. 

• Studying injury experience in two car collisions. This is cur- 

rently being tested for NHTSA under Contract NHTSA-7-3261. 

• Surveys of households or body shops to find incidence of low dam- 

age side impacts. The reliability of this method is low. 

We conclude that currently it appears most reliable to use towaway crashes 
as a basis for the analysis. 

In addition to obtaining a consistent sample of crashes, one has to obtain 
sufficient information about the crashes. Certain information is readily avail- 
able, such as make and model/year of the involved vehicles, and all associated 
characteristics. Age and sex of the occupants are also easily available as are 
impact areas on the vehicles. The velocities of the vehicles and the angle of 
impact, however, have to be reconstructed by fairly complex processes, which re- 
quire various assumptions about the characteristics of the vehicles involved. 
While not totally accurate, such results are still far superior to anything that 
could be derived from analysis of available mass accident data. 

The collection of new data should be biased towards low to medium severity 
side impacts, to help assure that the effects of side beams will be adequately 
sampled. Such accidents are most likely to be found at intersections in urban 
areas. In many studies, the question of whether the data are "nationally repre- 
sentative" is extensively discussed. For evaluation of side beam effectiveness, 
representativeness is not a problem; the effects of the Standard in specific 
crash situations can be estimated from a biased sample of crashes. Representa- 
tiveness becomes a problem only if one wants to estimate the effects of the Stan- 
dard relative to all crashes. To evaluate side beam effectiveness, it is better 
to obtain a biased sample from urban crashes, where most of the side impacts 
will be relatively minor, and side beams may be most effective. It is then pos- 
sible to correct for the bias and generalize the results to rural areas, where 
there are more high speed crashes in which side beams are apt to have little 
impact on intrusion and injury reduction because of the extreme severity of the 
crash effects. 



4-3 



FMVSS 215 



The purpose of FMVSS 215 is to prevent damage to safety related parts of 
cars in low speed crashes. In addition, it is expected that damage to other 
parts will also be reduced. 

The main problems with evaluating this Standard are: 

(1) It is very specific in terms of the vehicle parts and systems 

to be protected, and 

(2) It applies to low speed crashes, of which many are not report- 

able, and many of the reported ones are not investigated by 
the police or any other non-involved party. 

To obtain information on damage to the vehicle parts covered by the Stan- 
dard, at least the following approaches are potential candidates: 

(1) Identify and investigate in detail low damage crashes. 

(2) Analyze automobile insurance claims. 

(3) Analyze sales of repair parts for the protected vehicle parts 

and systems. 

(4) Analyze the frequency of towaway due to damage to the protected 

parts and systems. 

(5) Analyze the frequency of front (or rear) impacts relative to 

all impacts in old accident data, because damage reduction 
may bring certain collisions below the reporting threshold. 

The first approach encounters the second difficulty mentioned above: that 

low damage crashes are not reported. The question is: "How does one identify 

lowr speed ctashes? The leading possibility for identification suitable for 
statistical analysis is a survey of car owners. Even if the car owners respond 
to the survey, it is unlikely that more than rudimentary information on the 
crash can be obtained. To obtain details on vehicle damage, a followup vehicle 
inspection would be required. It appears highly doubtful that a sufficient num- 
ber of owners would agree to such inspection, if only because of the inconven- 
ience involved. Furthermore, the expense of inspection would be very high. 
Another problem is that a specific car owner might not be aware of no-damage 
collisions in which other drivers in their household have been involved with 
the car. 

The second approach — analysis of automobile insurance claims — is subject 
to the following problems: 

(1) Automobile insurance policy holders are a biased sample, by 

company policy, and by owner choice. Also, automobile insur- 
ance claims for low damage crashes are a self-selected sample. 

(2) The claims data automated by insurance companies are very limi- 

ted. To retrieve detailed data from the hard copy files is 
inherently difficult and likely to be prohibitively expensive. 



4-4 



(3) Two distinctly different kinds of insurance deal with vehicle 
damage: collision insurance and property damage liability. 

The first is limited to damage to the insured vehicle (and also 
to damage to other vehicles driven by the insured) , the second 
covers all property damage of third parties, including non- 
vehicle damage. In addition, the relation between claimant 
and insurance company in a liability case is adversary; there- 
fore, information availability may be limited. 

There appear to exist only two insurance data bases which are usable: 

Highway Loss Data Institute (HLDI) collision claim data, and detailed collision 
damage data sampled by State Farm Mutual Insurance Company. 

HLDI data contain the total amount of a collision claim, detailed car mo- 
del information, the applicable deductible, use of the car by a young driver, 
and rating area. Total claim figures are of extremely limited value: they re- 

flect the influence of collision types, of repair parts cost, and of repair 
labor cost, in addition to the influence of the physical damage. It appears 
impossible to draw any specific conclusions on damage reduction due to FMVSS 215 
from these data. 

State Farm Mutual Insurance Company has analyzed samples of collision 
claim repair bills beginning in 1973, Usually, these samples cover the current 
model year, but occasionally samples of all insured vehicles are made. For 
each case the damaged parts are identified. Comparing the frequencies of dam- 
age to certain parts between model years should allow a realistic estimate of 
changes in vehicle damage patterns. 

The third approach would analyze sales of repair parts, including parts 
which are protected by the Standard. Certain parts, e.g,, lenses to taillights, 
are model and model-year specific. Analyzing the time trends of sales of such 
parts in relation to parts not protected by the Standard could indicate an effect 
of the Standard. The main problems are: there are only a few parts which are 
model/model-year specific, and the manufacturer's sales records would have to be 
obtained. A statistical problem would be to account for fluctuating inventories 
held by distributors and dealers. Therefore, this approach appears to hold lit- 
tle promise. 

The fourth approach uses the fact that some of the parts protected are neces- 
sary for the operation of the vehicle, such as fuel system, cooling system, pro- 
pulsion system, steering and braking. If damage to them becomes less frequent, 
the need for towing crash-damaged cars should be reduced. Aside from the fact 
that towing is only indirectly related to the requirement of the Standard, this 
approach appears possible and promising. 

The fifth approach would use existing mass accident data, beginning with 
1972, and analyze the relative frequencies of front and rear impact accidents 
relative to all others. A reduction in damage might bring certain crashes below 
the reporting threshold and thereby reduce their relative frequency. Mass acci- 
dent data from Virginia and New York suggest that a change in reporting require- 
ments does indeed result in a change in actual reporting practice. Therefore, 
it is plausible that a reduction in damage will result in a reduction in reported 
accidents. An important advantage of this approach would be that it would analyze 
cars not satisfying the Standard when they were still new, and damage is mare 
likely to be reported. 



4-5 



With the exception of the analysis of State Farm data, the above approaches 
are speculative with the following two approaches having little promise. The 
analysis of sales of repair parts may encounter difficulty in data acquisition 
and is unlikely to provide much information, even if data could be acquired. 

The HLDI data for damage costs are so highly aggregated that there appears lit- 
tle chance of success using that base to determine the effectiveness of FMVSS 215. 

FMVSS 301 

The purpose of FMVSS 301 is to reduce deaths and injuries occurring from 
fires resulting from fuel spillage in motor vehicle accidents. The Standard at- 
tempts to achieve this goal through establishing limits to fuel spillage in ve- 
hicle test situations. 

The main problems with evaluating this Standard are: 

(1) The infrequency of fire-related deaths in fatal accidents. 

(2) Fires due to fuel spillage in accidents account for only a small 

percentage of vehicle fires, so that mass data bases with just 
motor vehicle fire data would be insufficient. 

(3) Due to pollution control requirements, considerable changes have 

been made to the fuel system, possibly increasing the fire hazard. 

Other problems in evaluating the Standard are: 

(4) Without special training and equipment, it is difficult to de- 

tect fuel spillage/fuel system rupture, in an onsite investiga- 
tion. 

(5) In the case of fires, and fire-related deaths, there is the ques- 

tion of the cause of death. And in multi-car accidents there is 
the question of which vehicle caused the fire. 

(6) Given the relatively low numbers of incidents of interest, the 

analyses will probably be limited to answering simple questions 
about whether there is any discernable effect of the Standard. 
Detailed analyses of makes and models or crash configuration may 
not be statistically meaningful, unless large effects actually 
exist . 

To obtain information on fire and/or fuel spillage, at least the following 
approaches are potential candidates: 

(1) Analyze the frequency of fuel system rupture in towaway accidents 

for various model years. 

(2) Analyze the frequency of fire and/or fuel spillage in motor vehicle 

accidents by using historical accident data from fire and police 
departments, or through new data collection. 

(3) Analyze the frequency of fire-related deaths in motor vehicle ac- 

cidents using various state Fatal Accident files and possibly 
Vital Statistics records. 



4-6 



Determining the frequency of fuel spillage in motor vehicle accidents will 
be difficult because of the fast evaporation rate of gasoline and other diffi- 
culties in detection. Also, until the 1977 model year, other vehicles (multi- 
purpose vehicles, vans, trucks, buses) up to 10,000 lb GVWR did not have to meet 
FMVSS 301, Therefore, these vehicles cannot be included in the basic analysis. 
However, these vehicles represent a significant portion of the vehicle popula- 
tion (20% of the passenger car sales in 1970, 29% in 1975) and any information 
gathered on them would be of value. 

The first approach encounters the basic problem of measuring fuel system 
rupture. The type of accident would have to be restricted to towaways in order 
to assure that the vehicle is available for thorough examination. The second 
approach reduces the stringency of the fuel system integrity question by focus- 
ing on visible evidence which is immediately observable and probably requires 
fire department attention. The information on fire/fuel spillage could be ob- 
tained from a variety of sources: (1) historical fire department records; (2) 

new data collection by police; and a limited number of cases from (3) the Na- 
tional Crash Severity Study (NCSS). The third approach , the study of motor ve- 
hicle fatalities due to fire, has the basic problem of sample size and data accu- 
mulation. Preliminary investigations indicate that four states* segregate fatal 
accident hardcopy files to make them readily accessible. We believe it is safe 
to infer that at least the majority of states also maintain easily accessed 
fatal accident files. 

In summary, the first approach is the most systematic and precise but it 
suffers from having relatively few early models in the accident population. When 
historical data are used, the second approach overcomes the first problem but 
encounters potential problems of data inconsistencies. If police collect new 
data, there is the time delay and underrepresentation of earlier models. How- 
ever, potentially more data could be made available. The last approach most 
directly addresses the objectives of the Standard. However, the infrequency of 
fatalities due to fires in motor vehicle accidents limits data availability. 

The evaluation of the effects of FMVSS 301 faces two potential problems: 

(1) The use of current information from specially investigated 

accidents implies that all cars preceding the Standard are 
"old." Therefore, deterioration of the fuel system — rust, 
corrosion, fatigue, deterioration of rubber or plastic com- 
ponents, etc. — may increase the risk of fuel spillage. 

(2) In older accident data, which involve pre-Standard cars when 

still "young" and presumably not (or less) affected by fuel 
system deterioration, it is not clear that fuel spillages and 
fires are reported completely or consistently. 

The degree to which these problems will arise is an empirical question which 
cannot be answered with the currently available information. It is quite likely, 
however, that they will have some effect. Therefore, it is not feasible to de- 
sign a straightforward evaluation plan which will result in the conclusion that 
FMVSS 301 has a specific effect of reducing fuel spillage by ILpercent, or that 

“ft 

Connecticut, North Carolina, and Texas have physically separate files. New 
York saves low file numbers for fatal accidents. 



4-7 



an effect, if any, is less than Y percent. One may possibly obtain such a re- 
sult, but it is quite likely that the only possible conclusion will be that there 
are other effects, possibly masking all or part of the effect of FMVSS 301. In 
such a case, only ad hoo analyses, designed to eliminate such effects as far as 
possible, promise some hope of isolating the effect of the Standard. 

Therefore, all approaches proposed above are to a large extent speculative. 
None will lead with certainty to a conclusive result. As a purely subjective 
judgment, it is expected that the analysis of new data to be collected will be 
the most promising approach, provided that there is no significant fuel system 
deterioration with age. The analysis of fire department records appears to be 
the second most promising example. Analysis of fatal accidents appears least 
promising by itself. Using any two, or all three of these approaches, however, 
may give convincing overall results because of the independent nature ot the 
basic data, even though each analysis by itself may be actually or potentially 
subject to uncontrolled influences. 

FMVSS 2Q8 

The purpose of FMVSS 208 is to reduce the number of deaths and overall sever- 
ity of injuries in motor vehicle accidents by establishing performance require- 
ments for the protection of vehicle occupants in crash situations. 

The principal difficulties in evaluating this Standard are: 

(1) The effectiveness of the existing implementation of the Stan- 

dard depends on the actual usage of the restraint system. 

Measures of such usage in actual accident situations are 
often based on estimates, 

(2) In meeting the Standard, an assortment of methods have been used; 

these must apply to a wide range of individuals and crash situ- 
ations . 

(3) Manufacturers can comply with the Standard under any of three 

options, and are continually encouraged to upgrade the effec- 
tiveness of their systems. 

Other problems in evaluating the Standard are; 

(4) The 1974 and some 1975 models had ignition interlocks which sub- 

stantially changed the degree of belt usage in those model year 
cars. 

(5) There are relatively few vehicles presently on the road meeting 

the more rigorous Option 2 criteria. However, recent agree- 
ments between DOT and the manufacturers promise to increase 
that number, but not before the 1980 model year. 

To obtain information on the effectiveness of this Standard, three. approaches 
have been proposed: 

(1) Analysis of a combined NCSS/RSEP* data base, 

(2) Analysis of accidents of existing air bag and passive belt 

vehicles with plans to incorporate new data. 

*RSEP - Restraint System Effectiveness Program;NCSS-Natlonal Crash Severity Study. 



4-8 



(3) Collection of a nationally representative sample of restraint 
system usage. 

The first two approaches concentrate on the effectiveness of the Standard, 
given the usage of the occupant protection system. The purpose of the third 
task is to provide the background necessary to determine the overall effect of 
the Standard in the entire driving population. 

Combining the RSEP and NCSS data bases will provide not only more data 
but also a broader range of model years and new information on impact speed.* 

The differences between the proposed analysis and the RSEP study lie in this 
newly available data. Tests can now be made for effects of speed, impact angle 
and possibly restraint system locking systems. The statistical analysis would 
also differ to a certain extent because continuous variables will be used, such 
as speed. 

In the case of passive systems, a limited number of air bag and passive 
belt-equipped vehicles are presently on the road — approximately 11,000 and 65,000 
respectively. Because of the limited numbers of vehicles made available with 
these options, the present population may be highly biased. However, the pre- 
sent agreement between DOT and the manufacturers promises to make these vehicles 
more broadly available — but for air bags not before the 1980 model year. There- 
fore, the analysis recommended in this case focuses on developing analysis pro- 
grams and some initial estimates of effectiveness, and then processing additional 
data as it becomes available. The recommended statistical analysis is very sim- 
ilar to that for the NCSS/RSEP data, to provide comparability of results. 

The restraint system usage survey is presented in response to a request 
expressed by the Contract Technical Monitor. The usage information obtained 
from existing accident studies is biased towards the accident population. 

Also, these studies rely largely on claimed system usage, although RSEP and 
other serious studies are very careful about this. The usage survey may reveal 
some differences between the general driving population and the usage in the 
accident population. 

In conclusion, the first analysis will address the additional questions 
about the effects of speed and angle of impact which could not be addressed in 
the RSEP study. The second analysis will concentrate on the passive systems 
and will prepare for the large number which will come into the vehicle popula- 
tion with the 1980 and 1981 model year cars. The third analysis is necessary 
to place the effectiveness of the Standard in an overall context. However, some 
may judge that existing restraint system usage studies already supply adequate 
information. 



BEV is being added to the RSEP data base; it was not available in the original 
study. 



4-9 



4.2 Integrating the Evaluation Approaches 



There are several reasons for integrating the evaluation approaches, e.g., 
multiple use of the same data base or other data collection techniques. Also, 
the evaluation approaches can be better sequenced to spread the level of effort 
and provide an orderly progression of preliminary and interim results. There- 
fore, it is suggested that the following occur at the same time: 

• Analysis of mass accident data for FMVSS 214, 215, and 301. 

• Analysis of NCSS/RSEP data for FMVSS 214 and 208. 

• Analysis of towaway data effort for FMVSS 214, 215 and 301. 

• Surveys for FMVSS 208 and 215. 

• Hardware cost data. 

In addition, there are some analyses which are relatively simple and straight- 
forward and should be done early in the evaluation: analysis of State Farm and 

HLDI data for FMVSS 215. Other analyses are distributed over the evaluation 
period because of the rate at which data become available (analysis of passive 
restraint systems) or probability of finding significant results (analysis of 
fire/fuel spillage data before analysis of fire-related fatalities). 

Although this integration of approaches offers a distinct potential for 
efficiency and cost savings, there will be some added burden in terms of plan- 
ning and coordination. Secondly, the combined analysis will be perforce less 
focused on any individual Standard. And, finally, it may be judged that cost 
effectiveness is not an important criterion and that comprehensiveness i^s, re- 
sulting in integration by Standard, rather than task similarity. 

4 . 3 Cost Data 



NHTSA has stated that to measure the consumer's out-of-poclcet expenses the 
cost categories should be: 

• Direct manufacturing t Manufacturers' markup 

• Indirect manufacturing • Dealers' markup 

• Capital investment (including testing) • Taxes* 

However, we feel that the consumer's initial costs are determined by a 
complex process, with different types of bargaining at the retail, wholesale, 
and manufacturing levels. It is well recognized, and also acknowledged by the 
auto manufacturers, that wholesale prices are set in response to market condi- 
tions, and that their relationship to manufacturing cost is loose. In a recent 
CEM 8tudy+ this question was examined and no relation was found between annual 
increases in manufacturers' cost of satisfying FMVSS 's as estimated by GAO, and 
the retail price increases. 

Certain cost categories can be estimated well: direct and indirect manu- 

facturing, and capital investment, including testing. These costs represent 
real resources used. The question of markups is conceptually very difficult, 
considering the manufacturers' pricing strategies (trying to cover a market 
spectrum) and the oligopolistic nature of the market. Using average gross 

*Personal communication from Warren G. LaHeist, Contract Technical Monitor, 

18 January 1977. 

^CEM Report 4194-574, Program Priority and Limitation Analysis, December 1976, 
Contract DOT-HS-5-01225 . 



4-10 



profits for the manufacturing markup would be incorrect and misleading. To 
find the true markup would require a major study examining manufacturers’ de- 
tailed cost data and pricing practices (internal and external). 

The question of dealer markup is somewhat easier to consider conceptually. 
However, to determine it in practice is complicated by the trade-in of used 
cars. It appears highly likely that there is no fixed percentage markup on 
the dealer level, but a more complicated relationship which depends on the value 
of the new vehicle, the trade-in and other market conditions. Using an average 
gross profit, or the difference between wholesale and retail prices, would also 
be inaccurate and misleading. 

With regard to the issue of taxes, this cost is not only borne in the form 
of a sales tax as the fraction of the components cost of the total car, but it 
is also accumulated at every stage of manufacturing in the form of property, 
payroll, sales (intermediate) and excise taxes. Income taxes are another cost; 
however, they are not directly related to the resources used but to the profit- 
ability of the manufacturers. 

Therefore, based on the above discussion, we consider it beyond the state- 
of-the-art to estimate the true out-of-pocket cost of new car buyers due to 
satisfying the FMVSS. Good estimates of the costs of real resources consumed 
can be made, but these costs apparently are not passed on immediately or directly 
to the consumer of that model. Other costs (markups and taxes) are conceptually 
and practically difficult to establish. The most reliable estimate of consumer 
cost would have to be aggregated over the entire market and a several year per- 
iod in order to account for changes in market strategy and conditions. 

Another point of concern with regard to the collection of data on cost 
items is the periods of comparison — one model year before the effective date vs. 
the model year that the Standard became effective or the next model year. The 
first point is that manufacturers have made changes to vehicles prior to the 
effective date of compliance, especially in the case of totally new models. 
Secondly, there is the learning curve effect in most manufacturing processes 
which will reduce the effective cost of manufacturing over time. With regard 
to this second effect, savings would be difficult to estimate, especially as 
these new components become more integrated into the basic structure of the 
vehicle. Therefore, using these time periods for comparison may tend to over- 
estimate the cost of the Standard. 

Generally, specific hardware costs will be collected for each Standard. 

The number of models for which costs will be collected depends on the differ- 
ences in costs and implementations between models and manufacturers — for side 
door beams fewer models need be sampled than for bumpers. For FMVSS 214, side 
door beams are considered; bumper systems for 215; fuel systems for 301; and 
restraint systems for 208. 



4-11 



5.0 METHODOLOGIES FOR EVALUATING THE STANDARDS 



5.1 Introduction 

The four FMVSS's which are the subject of this study apply to different 
motor vehicle systems and the performance criteria for each Standard vary con- 
siderably. Within this context, alternative evaluation methodologies have been 
proposed for each Standard, which vary in the anticipated value of their re- 
sults and the effort required to perform them. There are, however, two compo- 
nents of these diverse tasks which are similar and in some respects identical 
for otherwise unrelated analyses. These are the statistical techniques used 
and the associated data sources for each recommended analysis. This section 
will first describe the data bases currently (or imminently) available and 
other data sources needed for the various proposed analyses. Then a general 
description of the statistical methodologies employed will be presented, to- 
gether with a description of the detailed analyses proposed for each Standard. 
The final subsection will present a cost data acquisition plan to determine 
hardware costs for each Standard’s implementation. 

5.2 Sources of Data 



The objectives of the proposed analysis for each data source, both exist- 
ing bases and new data collection efforts are given in Table 5-1. A short 
description of each data source follows. 

Mass State Accident Data 

These are automated data files of reported motor vehicle accidents within 
a state. They are generally maintained by the State Motor Vehicle Department 
or State Police and are coded using police and occupant accident reports. The 
formats, information collected, means of access, and number of cases available 
vary considerably among states. Because of this, state accident files are 
not detailed enough for use in evaluating the Standards. In addition, a spe- 
cific mass data base might have sufficient information for one analysis but in- 
sufficient information for another analysis. Other problems with these data 
are questions of reliability and how completely reportable accidents are co- 
vered. Texas, North Carolina, and Hew York have extensive automated files which 
often have sufficient information for a proposed analysis. The particular state 
data bases suggested to be used and the expected number of cases available are 
described for each analysis. 

National Crash Severity Study (NCSS) 

The NCSS is an 18-month effort which began in October 1976 and will continue 
through March 1978. The goal is to collect data on 10,000 accidents by 1978. 

Data are being collected by seven NHTSA-sponsored organizations in eight loca- 
tions: Western New York (CALSPAN) , Michigan (HSRI) , Miami (Univ. of Miami), San 

Antonio, Texas (SWRI), thirteen other counties in Texas (SWRI), Kentucky (Univ. 
of Kentucky), Indiana (Indiana Univ.), and Los Angeles, California (Ultrasystems). 
The sampling criteria are based on towaway accidents which are divided into three 
strata. Stratum 1 is sampled at 100 percent and consists of accidents where an 
occupant’s injury requires at least an overnight stay in a hospital (includes 
fatalities) . Stratum 2 is sampled at 25 percent and consists of accidents where 
an occupant requires hospital attention but does not stay overnight. Stratum 3 
is sampled at 10 percent and covers all remaining towaways. 



5-1 



TABLE 5-1 

ANALYSIS OBJECTIVES FOR EACH DATA SOURCE 







Federal Motor Vehicle Safety Standards 


Data 


Bases 


FMVSS 214: 


FMVSS 215: 


FMVSS 301: 


FMVSS 208: ' 






Side Door 
Strength 


Exterior Protection 
(Bumpers) 


Fuel System 
Integrity 


Occupant Crash 
Protection 




Mass State 

Accident 

Data 


Preliminary analysis of 
vehicle age & other non- 
side beam related ef- 
fects, in preparation 
for NCSS data analysis. 


Analysis of frequency 
of vehicle damage by 
area of damage to de- 
tect shifts away from 
bumper areas. 


Analysis of proportion 
of fatal accidents in- 
volving fire in pre- 
us. post-Standard vehi- 
cles (pre-1975 data). 






NCSS 

(Towaway) 


Detailed analysis to 
evaluate effect of side 
beams on: 

• Occupant injury 

severity. 

• Passenger compartment 

intrusion. 






Primary analysis of 
the effectiveness of 
seat belts in reducing 
occupant injury. 


Automate;! 

Existing 

Data 


RSEP 

(Towaway) 








Same analysis as NCSS 
data after BEV has been 
added to RSEP file. 


Bases 


FARS 






Analysis of proportion 
of fatal accidents in- 
volving fire in pre- 
us. post-Standard ve- 
hicles (post-1974 
data). 






HLDI 




Analysis of distribu- 
tion of insurance 
claim payments between 
pre- vs. post-Standard 
vehicles. 








State Farm 

Insurance 

Data 




Analysis of damage re- 
pair data to compare 
bumper parts replaced 
in pre- us. post-Stan- 
dard vehicles. 






Non- 

Automated 

Existing 

Data 

Bases 


Fire/Police 

Department 

Data 






Analysis of proportion 
of all accidents invol- 
ving fire or fuel 
spillage in pre- us. 
post-Standard vehicles. 






Passive 

Restraint 

Tracking 

System 








Analysis of the effec- 
tiveness of passive 
restraints (air bag, 
passive belt) in re- 
ducing occupant inju- 
ry. (Existing data. ) 




Addi tional 
NCSS Type 
Data 

(Towaway) 


Supplement to NCSS data 
for data categories 
with insufficient cases 
to achieve desired 
levels of significance. 










Additional 

Passive 

Restraint 

Tracking 

System 








Analysis of the effec- 
tiveness of passive 
restraints in reducing 
occupant injury. 
(Future data. ) 


New 

D3ta 

Collection 


Restraint 
System 
Usage Survey 








Observations of occu- 
pant restraint system 
usage tabulated by 
occupant & vehicle 
stratifications. 




Bumper 
Accident 
Car Owner 
Survey 




Analysis of proportion 
of accidents which are 
low or no damage in 
pre- us. post-Stand- 
ard bumDer vehicles. 








Towaway 
Accident 
Survey 
(Tow truck 
operator 
Sites) 




Analysis of proportion 
of towaway accidents 
with frontal or rear 
involvement in pre- 
us. post-Standard 
bumper vehicles. 


Analysis of proportion 
of towaway accidents 
involving fuel system 
ruDture in pre- us. 
post-Standard vehicles. 





5-2 



Restraint Systems Evaluation Program (R3EP) 

The RSEP file contains data on 15,818 (weighted) occupants who were involved 
in towaway accidents of 1973-1975 model year vehicles in the calendar year 1974 
or 1975, Data were collected by five NHTSA-sponsored teams located in Western 
New York (CALSPAN) , Michigan (HSRI) , Miami (U. of Miami), San Antonio, Texas 
(SWRI) , and Los Angeles, California (USC) . The general sampling criteria were 
100 percent of all such accidents where at least one front seat occupant was 
treated by a hospital and 50 percent of all such accidents where no hospital 
treatment was involved. The latter data were chosen according to the odd-even 
status of the last license plate digit. There were variations to this scheme in 
specific sampling areas for specific time periods, but it was the primary scheme 
used. 

Fatal Accident Reporting System (FARS) 

NHTSA's FARS maintains detailed information on all fatal motor vehicle ac- 
cidents. It has been implemented beginning with 1975 accidents. Since FARS in- 
cludes data from all states, it is possible to use FARS instead of individual 
state fatal data from 1975 on. 

Highway Loss Data Institute (HLDI) 

HLDI is a non-profit organization that gathers, processes, and provides 
the public with insurance data. It has published a series of reports on colli- 
sion claims [l ]. The HLDI data contain the following information for make, 
series, and body type; 

• Insured vehicle years 

• Claim frequency per 100 insured vehicle years 

• Average loss payment per claim 

• Average loss payment per insured vehicle year. 

This information is given by deductible amount ($50 and $100) and operator age 
group (under 25, or not) and by model year and accident year. 

State Farm Insurance Data 

The State Farm data are a useful source of information with regard to dam- 
aged parts and their costs in collision claims. State Farm started collecting 
such damage repair estimates regularly for the current models in January 1973, 
as part of their "Current Model Year Study." At that time, similar information 
was also collected on selected 1972 vehicles. Some of these data were presented 
in Patterns of Automobile Crash Damage by Sorenson, Gardner and Cassassa [2 ], 
They also take occasional samples of all claims during a certain period covering 
all model years. 

Fire Department Data 

Fire departments throughout the country collect data on motor vehicle calls 
to which they responded. An example is given of the type of information avail- 
able at fire departments by describing the situation in Hartford, Connecticut. 

In Hartford from 1971 to 1976, the number of responses of the fire department to 
alarms ranged from 7700 to 13,800 annually. Each of these responses is entered 
on a single line of a log book with the reason for the alarm indicated. This 
log book can be scanned to determine which responses must be looked at in greater 
detail. During the 1971 to 1976 time period, the number of vehicle-related 



State Farm Mutual Automobile Insurance Company, Bloomington, Illinois. 



5-3 



responses ranged between 750 and 800 annually. The information which can typ- 
ically be derived from the detailed accident form is the following: (1) inci- 

dent number; (2) time and location; (3) vehicle year make, model, serial number; 

(4) vehicle occupants and injuries and fatalities; (5) occurrence of fire and/or 
fuel spillage; (6) location of fire and material ignited; (7) involvement in 
accident and single or multi-vehicle; and (8) type of collision (rear end, etc.). 

It is of considerable interest to note that, in Connecticut, state law requires 
a report to be filed by the Local Fire Marshall to the State Fire Marshall within 
10 days after each fire. Thus, reports contain the above information in summary 
form, together with a dollar estimate of damage. Thus, in Connecticut, all fire- 
related vehicle accident information from various cities and towns can be ob- 
tained at a single location (State Fire Marshall’s Office). Note: this is not 

true of fuel spillage accidents. Cross-tabulation with police department records 
may be necessary to acquire missing information. 

Passive Restraint Tracking Systems 

There are currently several sources which document air bag accidents. The 
NHTSA maintains a National Response Center which provides a 24-hour phone service 
for reporting air bag vehicle accidents. General Motors Corporation provides 
the National Response Center phone number on the sun visor of all its air bag- 
equipped cars. Once an air bag deployment is identified, NHTSA performs a Level 
2 or Level 3 accident investigation to record the relevant crash characteristics. 
Automobile insurance carriers are another source of information. Allstate In- 
surance offers premium discounts for air bag-equipped vehicles and believes it 
insures a high proportion of the existing air bag vehicle population. In addi- 
tion, Allstate operates its own fleet of approximately 475 air bag vehicles. 
Allstate also maintains its own 24-hour phone service for reporting air bag ac- 
cidents, and drivers in their fleet are instructed to report all accidents. 
Insurance claims on policies covering air bag-equipped cars are monitored, and 
the Chicago police cooperate by reporting any air bag deployments they encounter. 
Identified Allstate fleet accidents are investigated by Allstate, and all air 
bag crashes are reported to the NHTSA. Car manufacturers and other insurance 
companies also cooperate with Allstate in air bag vehicle accident reporting. 

There is currently only one passive belt implementation in actual produc- 
tion. This is the Volkswagen Rabbit passive shoulder belt system which has been 
an option since the 1975 model year. Volkswagen instructs its dealers to report 
Rabbit accidents to the main office when the damage cost is above a threshold 
quantity (approximately $700) and then sends out investigators to collect data 
on the accident. Volkswagen will then notify the Accident Investigation Divi- 
sion of NHTSA about the accident. This is the only accident tracking procedure 
known of for passive belts. 

The present plans are to manufacture 450,000 air bag-equipped automobiles 
in the 1980, 1981 model years. A more extensive tracking system must be designed 
to collect data on the future increased number of air bag vehicle accidents. 

Additional NCSS-type Towaway Accident Data 

The number of cases available from the NCSS data collection effort is not 
expected to be totally sufficient for the analysis of FMVSS 214. It is neces- 
sary, therefore, to collect additional accident data with a similar level of 
detail to obtain more cases in those categories which are underrepresented in 
NCSS. The initial analysis of NCSS will give a first estimate of the effective- 
ness of the Standard. Using this estimate and the desired confidence level, 
one can then determine the absolute number of additional cases required. If 
the effectiveness is greater in a speed range, or for some other set of conditions. 



5-4 



subsequent data collection could be explicitly targeted, thus requiring fewer 
observations. The new data collection sites should be the same eight areas as 
the earlier NCSS data collection — Western New York, Michigan, Miami, San Antonio, 
other areas in Texas, Kentucky, Indiana, and Los Angeles. The accidents of most 
concern will be urban and suburban, front-side collisions occurring at relatively 
low speeds. It is expected that the results of the initial NCSS data analysis 
will confirm this requirement. If the data collection effort lasts one year, 
an average of 375 to 625 cases per site will be required, 

Towaway Accident Data (Towtruck Operator Sites) 

Two proposed analyses, one for evaluating FMVSS 215 and one for FMVSS 301 
require data which would be collected at police-designated towtruck operator 
sites. For FMVSS 215, data will be collected with the cooperation of police- 
designated towtruck operators. The data will be collected over a period of a 
year at a sufficient number of locations to accumulate about 2000 bumper cases 
during that time period. The site could include NCSS data collection areas and 
also would preferably be located in states such as New York and Texas which 
have automated mass accident data bases. The following basic information on 
each towaway accident involving front and rear collisions is required: 

• Vehicle model year 

• Vehicle make/model 

• Reason for towing (to insure that an accident is involved) 

• Front /rear bumper involvement 

• Location of accident. 

In addition to the information for each front/rear towaway accident, a count is 
required of the total number of towaway accidents handled by the towtruck oper- 
ators . 

For FMVSS 301, more detailed information will be needed, requiring trained 
investigators. The fuel system components to be tested for rupture are: 



• 


Gasoline cap 


• 


Fuel pump 


• 


Filler pipe connector 


• 


Carburetor 


• 


Gasoline tank 


• 


Vapor control carbon canister 


• 


Fuel line and connectors 







The acquisition of fuel system rupture data in towaway accidents must ad- 
dress the following considerations: 

• Selection of sample regions, 

• Securing cooperation of police and police-designated towtruck operators. 

• Preparation of data forms and training of investigator/technician. 

• Requirements of sample size and length of study. 

Data will be collected with the cooperation of both the police and police- 
designated towtruck operators. The ability to secure such cooperation will in- 
fluence the selection of sample sites. It may be advantageous to locate the 
sample regions in National Crash Severity Study (NCSS) data collection areas. 



5-5 



Restraint System Usage Survey 

Estimates of restraint system usage are necessary if one wishes to project 
the total number of deaths and injuries avoided due to FMVSS 208. Previous 
studies of restraint usage have been done and this data collection would differ 
in the following ways: 



• Two-person teams to observe and record the information. 

« Broader range of highway types, including on-the-highway obser- 
vation and accompanying police on random roadside vehicle in- 
spection. 

• Collection of data in the same geographic areas as RSEP data: 

Western New York, Michigan, Miami, S; n Antonio, rural Texas, 
and Los Angeles. 

© Interview followups on a sample of observations to gain addi- 
tional information on trip type and length and consistency of 
belt usage and also to check overall data collection accuracy. 



The number of observations required depends on the desired accuracy of the 
estimate and the frequency of occurrence of the desired event. 

Car Owner Survey (Low Speed Accidents) 

The survey of vehicle owners is designed to collect data which will permit 
a study of cars with and without bumpers that meet the requirements of FMVSS 
215. Specifically, the analysis of data will be directed toward determining 
the frequency of collisions and the level of damage (including no-damage) at 
low speeds. The survey of car owners should be designed to determine informa- 
tion on vehicle accidents which occurred during the prior six months. The in- 
formation required for each accident is: 



The first two above items will be known and will be part of the basis for selec- 
ting the owner in the survey. The questionnaire must be clearly worded so that 
the respondent will realize that he or she is to include very minor collisions, 
such as "bumps" which resulted in little or no damage. 

The data acquisition, which is assumed to be undertaken by a company with 
survey data collection experience and competence, must address the following 
considerations : 



Survey data of the type required in this study could (at least in principle) 
be collected by either phone or mail. However, in our judgment, the amount of 
information required and the time for reflection on the part of the respondent 
that is needed to assure a valid answer, would dictate a mail survey. 



m Vehicle year 
« Vehicle make/model 
® Type of collision 



• Amount of damage, including none 

• Damage repaired or not 

• Towing of car required or not. 



® Means of survey data collection - mail and/or phone 

• Representative sampling 

• Sequence of sampling - pilot study 

b Response rates and sample size requirements. 



5-6 



5.3 Statistical Techniques 



The statistical techniques needed in the evaluation of the four Standards 
for each data source are displayed in Table 5-2. Descriptions of the applica- 
tions of each technique follow: 

Contingency Table Analysis 

Contingency table analysis is used for evaluating all four Standards, when- 
ever the attributes of the populations to be compared are categorical and the 
question of a significant difference between the two populations is under exam- 
ination. This corresponds in most cases to a comparison of pre- and post-Standard 
cars with respect to a related performance criterion (e.g., occupant injury). 

In the case of FMVSS 214, mass state accident data are to be analyzed, using 
contingency table analysis to determine if any significant vehicle age effects 
or other non-side-beam-related effects are present. The analysis procedure to 
be followed can be illustrated with reference to Table 5-3 and Figure 5-1. In 
this illustrative discussion, the factors of driver age and model year are being 
"controlled for"; all cases are limited to a given category. For example, the 
driver age category might be under 25 years old and model year could be 1970. 

In the notation in Table 5-3, m is a frequency count of drivers injured and 
n is count of drivers not injured.. In the instance of impact analysis, m is a 
frequency count of side impacts and n is a count of other impacts. The sub- 
script refers to the vehicle age, i.e., zero indicates less than one year old. 

The superscripts refer to the vehicle category and whether the struck car con- 
tained side beams. Thus, for example, Ajjg is vehicle category A without side 
beams. No weight subclassification was needed for Category A. The superscript 
lBg \refers to the first weight subcategory of vehicle Category B and side beams 
present in the struck vehicle. 

The cube shown in Figure 5-1 illustrates the fact that the accident data 
with and without side beams will be analyzed separately. For simplicity, only 
primary vehicle categories A through E are shown, without the weight subdivi- 
sions. Six categories of vehicle age are shown. For each cell in the cube, 
stratified according to side beams, a frequency count will be made of injured 
and uninjured drivers for a given vehicle age and vehicle category. 

The contingency table analysis will proceed as follows: Analyses will be 

performed separately for the side beam and non-side beam samples. Consider a 
given row of Table 5-3 for either side beams of no side beams. If there were 
no effect of vehicle category for a given vehicle age, it would be expected 



That is, the ratio of injured drivers to total drivers will not change signifi- 
cantly among vehicle categories. A comparison can be made of the observed and 
expected number of injuries in each category, where the expected number of in- 
juries is simply the proportion of injuries that would be expected if there were 
no effects among vehicle categories. For a given cell i, expected injuries 
are obtained from 



that 




5-7 






STATISTICAL TECHNIQUES USED IN ANALYSES 







Federal Motor Vehicle Safety Standards 


Data 


Bases 


FMVSS 214: 


FHVSS 215: 


FMVSS 301: 


FMVSS 208: 






Side Door 
Strength 


Exterior Protection 
(Bumpers) 


Fuel System 
Integrity 


Occupant Crash 
Protection 




State 

Accident 

Data 


9 Contingency Table 
Analysis 


a Contingency Table 
Analysi s 

9 Heuristic Analysis 


• Contingency Table 
Analysis 

- Likelihood Ratio 
Tests 


“ - 




NCSS 

(Towaway) 


a Regression Analysis, 
Including Analysis of 
Covariance 

• Log-linear Model 8 Con 
tingency Table Anal. 

e Index Analysis 
e Heuristic Analysis 






a Regression Analysis, 
Including Analysis of 
Covariance 

a Log-linear Model 8 Con- 
tingency Table Anal. 

a Index Analysis 
a Heuristic Analysis 


Automated 


RSEP 

(Towaway) 








(Same as above for 
NCSS data.) 


Existing 

Data 

Bases 


FARS 






a Contingency Table 
Analysis 

- Likelihood Ratio 
Tests 

e Trend Analysis 






HLDI 




• Comparison of 
Truncated Log- 
Normal Distributions 








State Farm 

Insurance 

Data 




9 Contingency Table 
Analysis 

a Heuristic Analysis 






Nan- 

Automated 
! Existing 
Data 
Bases 


Fire/Police 

Department 

Data 






a Contingency Table 
Analysis 

- Likelihood Ratio 
Tests 

a Heuristic Analysis 






Passive 

Restraint 

Tracking 

System 








(Same as above for 
NCSS 8 RSEP data.) 




Additional 
NCSS Type 
Data 

(Towaway) 


(Same as above for 
NCSS data.) 










Additional 

Passive 

Restraint 

Tracking 

System 








(Same as above for 
NCSS 8 RSEP data.) 


New 

Data 

Col lection 


Restraint 

System 

Usage Survey 








a Heuristic Analysis 




Bumper 
Accident 
Car Owner 
Survey 




a Contingency Table 
Analysis 

• Heuristic Analysis 








Towaway 
Accident 
Survey 
(Tow truck 
operator 
Sites) 




• Contingency Table 

Analysis 

• Heuristic Analysis 


a Contingency Table 
Analysis 

a Trend Analysis 





5-8 



TABLE 5-3 

SIMPLIFIED CONTINGENCY TABLE ANALYSIS 



Vehicle 

Age 


Vehicle Category 


A ns 


A s 


,8 ns 


18 s 


2B r,s 


2B S 




0 


A NS 

m o 




A s 

m o 


% s 


^ 'NS 
m o 


a* 


m 1B s 

m o 




2B ns 

m o 




2B S 

m 

0 




• • • • 


1 - 2 


a ns 

m l 


A NS 
n 1 


A s 

m 1 


A 


1B NS 
m 1 


n i 


!6 S 
m , 




m 2B NS 
m , 




7 s 

"i 




• • • • 


• 

• 

• 






























Figure 5-1 f Simplified three dimensional analysis of sidebeams 
presence, vehicle age, and vehicle category. 



5-9 



r • 



, where 



(n ± + n i ) 



row 

E m 

r = 

row 

E (m+n) 



The ratio r is the sum of the total driver injuries in the row divided by the 
sum of the total drivers involved in accidents in the row (i.e., for a given 
vehicle age) . The significance of the differences between the observed and ex- 
pected injuries (m^ - E-^) can be evaluated with a standard Chi-square test. 
Using the above procedure, the effects of vehicle categories on injuries can 
be evaluated for each vehicle age class. The identical analytical step as out- 
lined above will also be carried out in the evaluation of side impacts, where, 
in this case, m is the frequency count of side impacts and n is the count of 
other impacts. 



Using the same approach, an entirely analogous procedure can be undertaken 
to evaluate the effects of vehicle age. If there were no vehicle age effects, 
it would be expected that the ratio of injured drivers to total drivers would 
not change significantly among vehicle age categories within a given vehicle 
category column, 







where the subscripts 0, 1-2, and 3-4 indicate the definition of the first three 
age categories as given in Figure 5-1. The expected number of injuries Ej for 
a given cell j within a contingency table column illustrated in Table 5-3 would 
be : 

Ej = r^ • (mj + nj ) , where 
col 

■j Em 



E (m+n) 

Again, the significance of the differences between observed and expected injur- 
ies (mj - E^) can be evaluated with a standard Chi-square test. Thus, the ef- 
fects of vehicle age classes on injuries can be evaluated for each vehicle cate- 
gory. Collision impact effects can be similarly determined. 

The same type of contingency table analysis will be used with mass state 
data for evaluating FMVSS 215. The basic question to be answered is: 

© Has there been a shift in the distribution of vehicle damage 
away from bumper areas? 



5-10 



Answering the above question requires an analysis of the frequency of damage 
occurrence by area of vehicle. This can most appropriately be undertaken 
through contingency table analysis. The primary breakdown of area of damage 
would be front, side and rear. Where data permit, subcategorization of the 
damage area could be used. The analysis will attempt to determine if the fre- 
quency of reported accidents involving bumper systems has changed on new models 
since 1973 as compared with old models prior to 1973. This would be done to 
test for the underrepresentation of accidents involving bumpers which meet the 
requirements of FMVSS 215. If underrepresentation is the case, then it would 
support the hypothesis that the new bumpers are effective in reducing the dam- 
age to vehicles equipped with them. 

The comparative analysis of area damage frequency for pre- and post- 
Standard cars will require several data stratifications and controlling for 
extraneous effects. The shift (if any) in area damage frequency in the contin- 
gency table analysis may be more susceptible to detection if stratification 
according to damage severity is performed. It is possible that frequency shifts 
will be detected only in collisions with lesser damage. Additionally, it may 
be necessary to control for effects due to driver age and/or sex. For example, 
more younger persons drive older cars and, due to more aggressive driving char- 
acteristics , tend to be more frequently involved in front-end collisions. If 
this is the case, older (and predominately pre-Standard) cars could have a 
higher frequency of bumper- involved, accidents than newer (and predominately 
post-Standard) cars, but this effect should not be ascribed to the new bumper 
systems . 

The contingency table analysis should also be carried out for data strati- 
fied according to market class (subcompact, compact, intermediate, full size, 
heavy). The effects and effectiveness of the new bumper system may differ be- 
tween a subcompact and a full-size car. Additionally, there has been a shift 
in the relative market share of the above five vehicle classes in recent years, 
and this should be considered in the analysis. 

The analysis will initially be carried out separately by accident year. 
There are several exogeneous factors which might be changing over time. For 
example, a state may change the minimum dollar amount of damage required for 
an accident to be reportable. It has been observed in the past that when such 
reporting limits change, the number of accidents actually reported changes sig- 
nificantly. Exposure is another factor that changes over time. As the economic 
cycles change, the amount of driving changes correspondingly. If certain types 
of driving are affected more than others by the economy, the relative occurrence 
of different accident configurations may change. This would affect a comparison 
of frequency of accidents by damage area which combined all the accident years 
together. Depending on the results of the initial analysis, similar accident 
years may be combined to increase sample size, especially where accidents invol- 
ving pre-Standard vehicles are infrequent, as is the case with the latest acci- 
dent data. 

Contingency table analysis will also be the primary method used for anal- 
yzing other data sources for FMVSS 215. The analysis of State Farm Insurance 
data will compare the number of cars with bumpers replaced versus cars with 
non-protected parts replaced in pre- versus post-Standard cars (or any more de- 
tailed categorization) . The analysis of the Car Owner Survey will determine 
if post-Standard bumper cars are involved in a greater percentage of no-damage 



5-11 



or low damage accidents relative to all the accidents in which they are in- 
volved, than pre-Standard bumper cars. The analysis of data collected from 
towtruck operators will determine if vehicles with post-Standard bumpers have 
a smaller percentage of frontal or rear involvement in towaway accidents. An 
example of the corresponding contingency table is shown in Figure 5-2 below. 



Model 

Year 


Towaway Accident Impact 


Total 


Front/Rear 


Other 


< 1972 








>_ 1973 








Total 









Figure 5-2. Illustration of 2 x 2 contingency table analysis 
designed to estimate the reduction in front/ 
rear towaway accidents due to the effect of 
post-Standard bumpers (model year 1973 and 
later) . 

There are three separate analyses recommended for evaluating FMVSS 301. 
They are: 

• Analysis of Fuel System Rupture in Towaway Accidents 
© Analysis of the Frequency of Fire and Fuel Spillage 
» Analysis of Fire-related Fatal Automobile Accidents. 

Each uses contingency table analysis to compare pre- and post-Standard 301 
vehicles. The first analysis involves a 2 x 2 contingency table analysis with 
all cases in which obvious aging effects were not observed in the fuel system 
of the vehicle. The aging effects include pre-existing damage, corrosion, 
fatigue, crystallization of metal, extensive hardening of rubber or plastic, 
etc. The 2x2 contingency table analysis is outlined in Figure 5-3. A stan- 
dard test would be employed to determine if there is a significant differ- 
ence in the occurrence of fuel system rupture in pre-Standard V8. post-Standard 
cars . 



Model Year Class 


Fuel System Integrity 


Total 


Rupture 


No-Rupture 


Pre-Standard Cars 




» 




Post-Standard Cars 








Total 









Figure 5-3. Contingency Table Analysis for cars without obvious aging effects. 



5-12 



For the second analysis, contingency tables will be constructed according 
to the differences to be tested. The fundamental measures of the Standard's 
effectiveness are differences in the ratios of fire-related accidents to all 
accidents and fuel spillage accidents to all accidents for pre- V&V 3 U 8 post- 
Standard cars. The analysis will permit the examination of variations of this 
effect with calendar year, vehicle age and type of impact. Also possible dif- 
ferences as a function of location (state) may be identified. 

The third analysis will use mass state accident data for earlier years 
(pre-1975) and FARS data for subsequent years. A contingency table analysis 
will be performed according to the table in Figure 5-4 below. Hardcopy fatal 
files will be used to ascertain the occurrence of fire, which is not available 
on mass state files. The mass state files will be necessary for information 
on the non fire-related fatal accidents. 





Fire-Related 


Non-Fire-Related 




Fatal Accidents 


Fatal Accidents 


Pre-Standard 








Vehicles 








Post-Standard 








Vehicles 









Figure 5-4. Contingency table for analysis of fire-related 
fatal accidents. 



Analysis of Covariance (ANACOVA) 

This method of analysis will be used in evaluating FMVSS 214 and FMVSS 208, 
For each of the two Standards a multinomial response model has been proposed 
with both continuous and discrete explanatory variables. Since the model in- 
volves a quantitative or regression component and a qualitative or analysis of 
variance component, the most plausible approach seems to be to consider the 
setup as an analysis of covariance problem. In using such an approach, the 
regression portion of the model (i.e., the continuous variables) is fitted by 
estimating the coefficients of the continuous variables. Then the analysis of 
variance portion of the model (i.e., the discrete variables) is considered in 
the presence of these covariates. Package programs are available to handle an 
ANACOVA of the size we are discussing so that "in principle" the analysis may 
be performed. Included in these packages are provisions to run significance 
tests and to obtain confidence intervals for the regression coefficients and 
also to run significance tests and multiple comparisons for the main and inter- 
action effects. This is the most promising approach for evaluating the effec- 
tiveness of side beams in reducing "extent of intrusion," 

Analysis of covariance is not as promising an approach when using injury 
severity (AIS) as the dependent variable as opposed to "extent of intrusion." 

The problem stems from the fact that ANACOVA assumes the dependent variable to 
be continuous and normally distributed. This assumption is not valid for the 
AIS scale. There are other problems of interpretation in using ANACOVA in this 
case.* An alternative analysis for using injury severity as the dependent 

See discussion on page 4-24 of Task 4 and 5 Report [ 3 ] . 



5-13 



variable is given in the "log-linear" analysis section. The models proposed 
for FMVSS 214 and FMVSS 208 are displayed in Figures 5-5 and 5-6, respectively. 



Continuous Mean Effects : 

• Impacting Speed of the Striking Vehicle 

- Denoted by S and enters quadratical ly 
@ Change in Velocity 

- Denoted by aV and enters quadratical ly 
9 Angle of Impact 

- Denoted by a and enters trigonometrically 

* 



Discrete Mean Effects: 



@ Seatbelt Status: 


B - Categorical 




@ Model Year Group: 


M - Dichotomous 




@ Occupant Age: 


A - Categorical 




© Presence of Adjacent Occupant: 
Recommended Model : 


J - Dichotomous 




Log p = u 


2 

+ a -j A V + a 2 AV + b i S + 


f 2 S 2 


(continuous) 


+ B, + Aj + M k * 




(categorical) 


+ c^AV sina + c 2 aV sin2a + 

9 9 


2 

C^AV cosa C^AV sina 

p 


(continuous 

interactions) 


+ CgA V sin2a + CgAV COSa 


+ d^S sina + d^S sina 




+ B. COSa + M cosa + M.‘ 

i k k 


aV + J ? sina 


(continuous 

categorical 

interactions) 


★ 

The variable list is only illustrati 


ve in that the specific 




variables included will change as the analysis progresses. 





Figure 5-5. Multinomial response model for FMVSS 214. 



5-14 



Variable 


Type 


Definition 


A V = Change in Velocity 
I = Impact Point Angle 
A = Force Angle 
W = Weight of Case Vehicle 

M = Model Year Group 

G = Age of Occupant 
S = Sideswipe Variable 


Quadratic 

Angular 

Angular 

Nominal 

Dichotomous 

Nominal 

Dichotomous 


NCSS file definition 

See Figure 3-1, Reference [6] 

See Figure 3-1, Reference [6] 

Weight categories < 2000 lb, 2000-3000, etc. 

Model Year categories: before 1969, after 1969 

Age groups 16-25, 26-35, etc. 

No Sideswipe = 0, Sideswipe = 1 



Recommended Model : 
Log p = u 



+ 


a l V 


+ 


a 2 


V 2 


















(continuous) 


+ 


w i 


t M. 

J 


h G k + 


















(categorical) 
























*\ 






+ 


b-j AV 


cos 


I 


+ b 2 AV 


cos 


21 


+ 


b 3 AV 


cos 


31 
































> 


(continuous 


+ 


CjAV 


sin 


I 


+ c 2 AV 


sin 


21 


+ 


d-j AV 


cos 


A 




interactions) 


+ 


d 2 AV 


cos 


2A 


+ e,AV 


sin 


A 


+ 


f^V 


cos 


(A+I ) J 







where p is the probability of equaling or exceeding a particular AIS level 
for a particular belt system usage, and 

p, d-| , f-j 

are coefficients to be estimated from the data. 



Figure 5-6. Multinomial response model for FMVSS 208. 



Log-linear Analysis 

This technique is an alternative to the ANACOVA analysis described above. 

It will also be used to evaluate FMVSS 214 and FMVSS 208. It is a preferable 
approach for using injury severity as the dependent variable because it retains 
the multinomial character of the dependent variable at a relatively minor sacrifice. 
If categorization is imposed onAV and Angles in the models in Figures 5-5 
and 5-6, then a log-linear model may be fitted to the data. The log-linear 
model assumes a higher order contingency table type categorization with respect 
to the observed independent variables and &. dichotomous response for the depen- 
dent variable. The logarithm of the probability of one of these responses is 
given a linear representation in terms of the levels (categories) of the inde- 
pendent variables. The model then only requires that at a given set of levels 
for these variables, observed responses follow a binomial model with the cor- 
responding model-specified probability of occurrence. The model we have given 
need only be amended with respect to the continuous portion. 



5-15 



In the evaluation of FMVSS 214, log-linear analysis will be used to detect 
differences in the probability of occupant injury less than or equal to a given 
severity, in side beam versus non-side beam cars. The data used will be exist- 
ing NCSS data and any additional NCSS-type data obtained in a new data collec- 
tion effort. In the evaluation of FMVSS 208, this analysis will be used to 
detect differences in the probability of occupant injury less than or equal to 
a given severity, as a function of the restraint used. In seat belt equipped 
vehicles this results in the following stratifications: 

• Lap belt only used 
© Lap/shoulder belt used 
» No restraint used. 

In air bag equipped vehicles the stratifications will be: 

© Air bag deployment with lap belt used 
® Air bag deployment without lap belt used 
® Air bag non-deployment with lap belt used 
e Air bag non-deployment without lap belt used. 

In passive belt equipped vehicles the stratifications will be: 

® Passive belt used 
9 Passive belt not used. 

NCSS data, RSEP data and Passive Tracking System data will be included in these 
analyses. A flow chart of the proposed analysis schemes appears in Figure 5-7 
at the end of this subsection.* 

Index Analysis 

A third procedure has been proposed to compare the protection afforded by 
the three categories of seat belt usage. Let P^ denote the probability of in- 
jury at least as severe as AIS = 3 (i.e., AIS ^_3) when the driver is not using 
seat belts. Let p£ and Pg be the corresponding probabilities with lap belts 
and shoulder/lap belts, respectively. We propose the index 

P 3 

3 N 

I (L,N) = log -y 

P L 

as a measure of the improved protection of lap belts over no belts for AIS >_ 3. 
For other injury levels the definition is similar. This index has several de- 
sirable properties. If the probability of injury is the same, P^ = P^, then 
I 3 (L,N) = 0. Should lap belts decrease the probability by 1/2, then' P^ = 1/2 P^ 
and 

I 3 (L,N) = log 2 2=1. 



Flowcharts of other selected analyses are presented in Appendix I. 

The choice of the base for the logarithm is arbitrary. Base 2 was chosen be- 
cause it is conceptually desirable for differences on the order of 0.5, e.g., 
between belts and no belts. Log e wou ld be conceptually more desirable for 

small differences because it would correspond to percentage differences. Pre- 
ference in choice of base for the logarithm can be investigated further when 
performing the analysis. 



5-16 



Conversely, if no use of belts decreases the probability by 1/2, then = 
1/2 P 3 , and 

I 3 (L,N) = log 2 l/2 = -1. 



Furthermore, the index is additive in the following sense. If I 3 (L,N) =1.8 
and I 3 (S,L) = 0.5, then 



I 3 (S,N) = 2.3. 

Also, note that order is important: I 3 (L,N) = -I 3 (N,L). 

Since the estimates of the injury probabilities are functions of the inde- 
pendent variables, the indices are also functions of these variables. This is 
desirable because any improvement due to seat belts would not be expected to be 
uniform across all situations. 

Trend Analysis 

The physical condition of a vehicle's fuel system prior to an accident 
will affect the probability of that system's rupturing in a collision. The 
analysis of FMVSS 301 will attempt to isolate vehicles with serious aging ef- 
fects which are defined as a pre-existing condition of the fuel system that 
would greatly increase the likelihood of rupture. Two separate trend analyses 
will be performed with data from towed vehicles. The first step consists of a 
relatively simple analysis of the frequency of occurrence of observable aging 
effects by model year. Obviously, the entire sample of cars with and without 
aging effects is to be utilized. The analysis is designed to identify discon- 
tinuities and/or changes in the trend of the occurrence of obvious aging effects 
of fuel system components by car age (i.e., model year). The detection of such 
an effect, if relatable to the Standard, could indicate that improvements in 
the materials used to comply with the Standard have reduced the aging effects 
of corrosion, fatigue, etc. 

The second analysis is a trend analysis of the occurrence of fuel system 
rupture in cases with significant observable aging effects. The trend analysis 
is designed to identify discontinuities and/or changes of slope in the trend of 
rupture (by model year) in accidents where there are obvious aging effects in 
the fuel system components. 

Likelihood Ratio Tests 

If the trend analyses described above discern a significant age effect in 
the likelihood of fuel system rupture, the nature of that effect must be consi- 
dered in the subsequent analyses of FMVSS 301. If the age effect can be real- 
istically divided into two or three discrete categories, then a standard contin- 
gency table analysis can be used. If, however, vehicle age must be included 
as a continuous linear variable, then a likelihood ratio test should replace 
the contingency table analysis. 



5-17 



Comparison of Truncated Lo^ Normal Distributions 

This analysis approach for FMVSS 215 is intended for Highway Loss Data 
Institute (HLDI) repair cost data to determine whether repair cost distributions 
differ between pre- and post-Standard bumper vehicles. Two methods are described 
which differ in that the first develops statistical estimates of the character 
of the truncated distributions and compares these estimates. The second com- 
pares the distributions within intervals. This latter method is the more power- 
ful, given large sample sizes. It is appropriate to note here that success in 
delineating the effectiveness of FMVSS 215 by either of these methods is spec- 
ulative. 

Outline of Approach 1 ; Suppose each of two sets of samples is taken from 
a truncated log normal distribution. The assumption of a functional form for 
the distribution enables estimation (maximum likelihood or method of moments) 
of the parameters of each distribution. However, the development of a test 
statistic for the comparison of samples must be ad hoc because of the absence 
of a large sample distribution theory for these estimators. This approach is 
preferred for estimation of parameters. 

Outline of Approach 2 : Suppose the samples are censored — that is, for the 

ith population (i = 1, 2), a total of observations (accidents) is taken, but 
only MjL are uncensored (i.e., actual repair costs are observed and the re- 
mainder are censored by the current value of the car) . This corresponds to de- 
veloping tests based on the first order statistics from the first sample and 
the first M 2 order statistics from the second sample. Nonparametric procedures 
using Generalized Wilcoxon test statistics are available to compare the popula- 
tion under this arrangement, and these test statistics are known to be asymp- 
totically normal. Since no functional form is specified, estimation must be 
confined to percentiles (i.e., medians, quartiles, etc,). This approach is in- 
tended to test the hypothesis of no difference between repair cost distributions 
for pre-Standard and post-Standard cars. 

Heuristic Analyses 

Heuristic analyses as described in this report refer to non-rigorous tabu- 
lations of available data to help the analysts decide which alternatives are 
the most promising as the research progresses. This could include simple tests 
of data homogeneity or stratified tabulations of the data to determine how many 
sample points fall into each category. 

One important such analysis will be done with the results of the restraint 
system usage survey. The analysis of the restraint system usage data would be 
rudimentary, primarily examining various patterns of usage through different 
tabulations. The tabulations of most interest will be seat belt usage versus: 

• Age • Restraint system 

• Sex • Vehicle class. 

• Rural /urban 

and possibly combinations of these with other variables. Simple tests of inde- 
pendence should be made to determine whether estimates are significantly dif- 
ferent from one another. 

The main questions addressed will be whether this study (1) finds any dif- 
ference from earlier studies and (2) finds substantial differences between cat- 
egories which had not been established before, such as rural/urban usage, or 
by trip type. 



5-18 



1.0 






. Some Continuous \ 
I Variables ; Others I 
\ are Categorical I 



X 



2.0 



Analysis of 
Covariance Model 



3.1 



3.2 



I 



Detailed Tabulation 
of Data, & Plots of 
Selected Scatter Diagrams 



3.7 



Analyze Tabulated Data 
and Scatter Diagrams 



3.3 



Iteration 



Select Variables 
Considered to be 
of Likely Importance 



3-A 



Test Analysis-of- 
Covariance Fit 



3.5 



Review Results for 
Goodness of Fit, 2 
by Sum-of-(Errorsr 
or Explanation of 
Variance 



No 




Detailed 

Accident 

Data (RSEP.’ICSS) 



Gross Tabulations: 

• Ferform Simple Tests 
Of Data Homogeneity 

« Separate Tabulations: 

AIS, Seating Posi- 
tion, Impact Site, 
Collision Type, etc. 



Fitting of 
Statistical Models 

• For AIS Levels 

FMVSS 214 

• For Body Area 

• For Intrusion Levels 

FMVSS 208 

• For Restraint Usage 

• For Seating Position 



Exogenous 

Information 

(e.g., engineering 
aspects, etc.) 



/ atz \ 

I Categorical I 
\ Variables J 






Log-Linear Model 



a.i 



Detailed Tabulation 
of Data 



A. 2 



4.3 



Select Variables 
Considered to be 
of Likely Importance 



A. A 



Test Log-Linear 
Model Fit 



A. 5 



Review Results for 
Goodness of Fit, 
by X? Analysis 



Preliminary Analysis 
Phase 



Analyze Data Cells: 
Perform Tentative 
Classifications 



A. 7 



Iteration 




® 



(see next page) 






Figure 5-7. Statistical Analysis Scheme for evaluating FMVSS 214 and 
FMVSS 208. 5-19 



r 






Deta i 1 ed Ana lys i s 
Phase 







Figure 5-7. (Continued) 



5-20 








5.4 Hardware Cost Data Acquisition 



This subsection presents a plan to collect hardware costs on vehicle com- 
ponents which are affected by the Standards. It will consider only components 
which are directly affected, not associated design changes. Each Standard will 
be discussed separately, first with a description of relevant cost items and 
then a suggested cost sampling plan. The sampling plans give examples of spe- 
cific car models to sample which are based on the particular components in- 
volved. The examples presented are intended as a descriptive device rather 
than a formal recommendation. After specific manufacturer/make/model to sam- 
ple have been decided upon for each Standard, the four plans could be combined 
into one integrated cost sampling plan. This last task is beyond the scope of 
this study. More detailed descriptions of the cost sampling plans may be found 
in each Standard's Task 4 and 5 report [3], [4], [5], [6], 

5.4.1 FMVSS 214 

FMVSS 214 was introduced in October 1970 with an effective date of January 
1, 1973. Manufacturers had been installing side door guard rails in some cars 
since the 1969 model year. Figure 5-8 shows the incremental design changes 
used to meet the Standard. .The door beams are approximately eight inches high, 
two inches deep and run from hinge to lock pillar on every door. They are par- 
allel to and approximately 10 inches above the lower door sill. The pillar 

! support is for local reinforcement for the door pillar. Therefore, the two 
primary physical items which are introduced to satisfy the Standard are the 
side beams and the pillar supports . The side beams themselves are made up of 
several components. The minimum components are the channel beam and the end 
plates. Domestic models have corrugated sheet metal for additional reinforcing 
and in vehicles with wide doors a center plate may be added. The pillar to floor 
reinforcement is not required on 2-door sedans. 




Figure 5-8. Sketches of design changes required for FMVSS 214. 



*Source : Benefit and Cost Analysis Methodology . Reference [7]. 



Since side door guard beams are the universal method of compliance through- 
out the industry, cost variations among manufacturers should be less for FMVSS 
214 than for the other three Standards this project will review. We do expect 
real differences according to body styles and car classes. For example, the 
cost of four short beams in a 4-door sedan should differ from the two longer 
beams in a 2-door hardtop. Similarly, we expect the cost of a large luxury 
car's side beam to differ from a subcompact's side beam. For these reasons, we 
propose a three-dimensional categorization for cost data acquisition. 

Exhaustive Cost Acquisition Plan : 

1. Manufacturer: GM, Ford, Chrysler, AMC, Volkswagen, Toyota. 

2. Market Class: Subcompact, Compact, Intermediate, Full Size, 

Luxury, Specialty. 

3. Body Type: 2-Door Hardtop, 2-Door Sedan, 4-Door Hardtop, 

4-Door Sedan, 2-Door Hatchback, 

4-Door Station Wagon.* 

A sample Latin Square Design is given in Figure 5-9 below for analyzing cost 
data. 



Manufacturer 


Sub- 

comp. 


Com- 

pact 


Inter- 

med. 


Full 

Size 


Lux- 

ury 


Spec- 

ialty 


Note : 

A, B , . . . F 
represent 
body 
styles 


GM 

Ford 

Chrysler 

AMC 

Vol kswagen 
Toyota 


A 


B 


C 


D 


E 


F 


B 


F 


D 


C 


A 


E 


C 


D 


E 


F 


B 


A 


D 


A 


F 


E 


C 


B 


E 


C 


A 


B 


F 


D 


F 


E 


B 


A 


D 


C 



Figure 5-9. Sample Latin Square Design for FMVSS 214 cost data acquisition. 

5.4.2 FMVSS 215 



The relevant cost items affected by FMVSS 215 are: 



o Front Bumper System : 

License Plate Bracket 
Bumper Guards with Protective 
Strips 
Face Bar 

Face Bar Impact Strip 
Face Bar Reinforcement 
Energy Absorbers 



Bumper Spring Assembly 
Filler Panel 
Frame Mounting Brackets 
Bumper Valance 
Air Deflector 

Brackets, Braces, Insulators, 
Sight Shields, Spacers 



& 

Additional investigation may show whether this classification can be further 
aggregated. 



5-22 



• Rear Bumper System : 

License Bracket 
Bumper Guards with Pads 
Face Bar Protective Strip 
Face Bar 

Face Bar Reinforcement 
In the case of the soft-face bumper 

are; 



Energy Absorbers 
Frame Mounting Brackets 
Filler or Valance Panel 
Heat Shield 

Brackets, spacers, etc. 
system, the components front and rear 



• Fascia skin 

• Elastomeric energy absorbers 

• Steel backing beam. 

Manufacturers will generally use the same bumper construction for all their 
c «*t lines, although there may be changes from year to year. There do exist sig— 
nificsnt implementation differences among manufacturers. These differences will 
increase the variance of estimates for the cost of complying with FMVSS 215. 
Although the individual manufacturer will use the same bumper construction on 
virtually all models, the cost will vary with car size. We, therefore, propose 
that cost data be stratified by market class and manufacturers, as follows: 

1. Manufacturer: GM, Ford, Chrysler, AMC, VW, Datsun. 

2. Market Class: Subcompact, Compact, Intermediate, 

Full Size, Luxury, Specialty. 

The recommended experimental design is shown in Table 5-4 below. 



TABLE 5-4 

SAMPLE EXPERIMENTAL DESIGN FOR FMVSS 215 COST DATA ACQUISITION 



Market Class 


Replication 1 


Replication 2 




Subcompact 


VW 


GM 


Compact 


Chrysler 


GM 


Intermedi ate 


GM 


AMC 


Full Size 


Ford 


Chrysler 


Luxury 


GM 


Ford 


Specialty 


Ford 


Datsun 






5-23 



5.4.3 FMVSS 301 



The vehicle components which are a part of the fuel system, and thereby 
affected by FMVSS 301, are listed in Table 5-5 below. Costs relating to chan- 
ges in these items which were made as a result of FMVSS 301 should be included. 

TABLE 5-5 

VEHICLE COMPONENTS AFFECTED BY FMVSS 301 



Fuel Tank 
Fuel Tank Filler 
Fuel Fi 1 ler Cap 

Fuel Tank Connection with Fuel and Vent Lines 
Fuel Tank Straps and Anchor Points 
Fuel Line 

Fuel Line Connections 
Vent Line 

Vent Line Connections 



Carburetor 
Fuel Pump 
Fuel Filter 

Connections and Mountings 



Automobile fuel system configurations vary considerably among manufacturers, 
makes, and model years. The Standard specifies maximum allowable leakage in a 
crash without defining specifications for particular fuel system components. 
Therefore, each manufacturer may or may not have changed various vehicle compo- 
nents as a result of FMVSS 301. This would make it very expensive and ineffi- 
cient to collect cost data on each fuel system component. Fuel system cost 
data should be acquired from manufacturers stratified by market class, but in 
the aggregate for the model's complete fuel system. The recommended experimen- 
tal design with a sample allocation of manufacturers to market classes is shown 
in Table 5-6 below. 



TABLE 5-6 

SAMPLE EXPERIMENTAL DESIGN FOR 
FMVSS 301 COST DATA ACQUISITION 



Market Class 


Replication 1 


Replication 2 


Subcompact 


VW 


GM 


Compact 


Chrysler 


Ford 


Intermediate 


AMC 


GM 


Full Size 


Ford 


Chrysler 


Luxury 


GM 


Mercedes 


Specialty 


GM 


Ford 


Mul tipurpose 


Chrysler 


GM 



* 

Fuel system costs ••inn modifications for multipurpose vehicles may be significantly 
different from passenger cars. 



5-24 



5.4.4 FMVSS 208 



The major components of the active and passive belt systems and the pas- 
sive air cushion system are summarized in Table 5-7 below. Costs relating to 
these items should be included. 



TABLE 5-7 

MAJOR COMPONENTS OF COMPLIANCE APPROACHES TO FMVSS 208 

Passive Air Cushion Approach [8,9] 

Driver air cushion and inflator assembly 

Passenger air cushion 

Air tank and inflator assembly 

Driver and passenqer knee restraints 

Dashboard indicator warning light 

Dashboard sensor 

Front bumper detector 

Lap belts at all designated seat positions 
Lap belt anchors 



Passive Upper Torso Belt Approach [10] 
Knee restrainer panel 

Single upper torso belt in front outboard positions 

Automatic belt retractor 

Floor anchors for belt retractor 

Seat belt warning system 

Reinforced anchorage on side doors for upper torso belts 
Lap belts for designated rear seat positions 
Rear seat belt anchors 



Active Type 2 Lap/Shoulder Belt Approach [11,12] 
Seat belt warning system 

Two 3-point lap/shoulder belts for front outboard positions 

Lap belts for other designated seating positions 

Shoulder harness retractors 

Lap belt retractors 

Floor anchors for retractors and belts 



FMVSS 208 has changed through the years and manufacturers’ methods of com- 
pliance have changed in response. For cost data acquisition for active systems, 
we are concerned only with implementations that are currently in production, 
which eliminates from consideration all but the three-point combination lap/ 
shoulder belt for outboard front seat occupants. Within each manufacturer there 
are three safety belt configurations, depending on the size of the vehicle: 

• Four seater - 2 lap/shoulder belts in front 

2 lap belts in rear 

• Five seater - 2 lap/shoulder belts in front 

3 lap belts in rear 

• Six seater - 2 lap/shoulder belts (outboard) ;1 lap-belt (center) in 

front 

3 lap belts in rear. 



5-25 



All the current lap/shoulder belts in production use one or both of the follow- 
ing inertia activated systems: 

• Mechanical locking activated by electronic vehicle decelera- 

tion sensor. 

• Totally mechanical locking activated by sudden pulling action 

on belt. 

We will assume for cost purposes that all manufacturers use basically the same 
locking retractor system for lap belts. The experimental design shown in Table 
5-8 is a balanced incomplete block design which is also balanced for the effect 
of inertia reel system. 

Manufacturers I to TV are the four major U.S. companies: GM, Ford, Chrysler 

and AMC. Manufacturers V and VI are foreign companies chosen on the basis of 
volume or possibly a unique restraint system. The assignment of manufacturers 
to specific columns is arbitrary and may be rearranged according to appropriate 
car production configurations. For those manufacturers which use only one type 
of inertia reel, both cost entries may be taken from the corresponding configu- 
ration type. For example, if Manufacturer I uses only inertia system "A," both 
4 seat and 5 seat costs may be entered using "A” system costs. If a manufacturer 
produces more than one model with identical seating configurations and the re- 
straint system costs differ, the model with the largest sales volume may be 
chosen. 



TABLE 5-8 

BALANCED INCOMPLETE BLOCK DESIGN FOR SAFETY BELT 
COST DATA ACQUISITION 



Confi gurati on 


Manufact 


:urer 


I 


II 


III 


IV 


V 


VI 


4 Seats 


A 






B 


A 


B 


5 Seats 


B 


A 


B 


A 






6 Seats 




B 


A 




B 


A 



o A = Electrically activated inertia reel. 
• B = Mechanically activated inertia reel. 



The cost data acquisition plan in Table 5-8 is only intended for implemen- 
tations that fall into FMVSS 208 - Option 3. There are only two current im- 
plementations which fall into Option 2. The Volkswagen Rabbit passive belt 
and the General Motors ACRS air bag/lap belt system. Both are unique enough 
to justify separate cost data acquisition and analysis. 



5-26 



5 ,5 References for Section 5 



1. Highway Loss Data Institute, Automobile Insurance Losses Collision Cover- 

ages 3 Variations by Make and Series 3 1973 Models 3 1974. 

2. Sorenson, W. W. , R. E. Gardner, and J. Casassa, "Patterns of Automobile 

Crash Damage," Automotive Engineering Congress 3 Detroit, Michigan, 

February 25-March 1, 1974. SAE 740065 

3. CEM Report 4207-564: "Final Design and Implementation Plan for Evaluating 

the Effectiveness of FMVSS 214: Side Door Strength, "Tasks #4 & //5 Report 

(Report #6), January 1977. 

4. CEM Report 4207-565: "Final Design and Implementation Plan for Evaluating 

the Effectiveness of FMVSS 215: Exterior Protection," Tasks //4 & #5 

Report (Report # 7) , February 1977. 

5. CEM Report 4207-566: "Final Design and Implementation Plan for Evaluating 

the Effectiveness of FMVSS 301: Fuel System Integrity," Tasks #4 & #5 

Report (Report //8), February 1977. 

6. CEM Report 4207-567: "Final Design and Implementation Plan for Evaluating 

the Effectiveness of FMVSS 208.: Occupant Crash Protection," Tasks #4 

and #5 Report (Report #9) , March 1977. 

7. National Highway Traffic Safety Administration, Benefit and Cost Analysis 

Methodology - MVP Rulemaking Programs 3 Technical Report, August 1972. 

8. Anonymous, 1973 Report on Progress in Areas of Public Concern 3 General 

Motors Technical Center, February 1973, 

9. Anonymous, Automotive Air Bags 3 Questions and Answers 3 Allstate Insurance 

Company, July 1976. 

10. Cooke, C., Consumer Cost of Occupant Crash Protection Sy stems , Motor Vehicle 

Programs, National Highway Traffic Safety Administration, July 1976. 

11. Anonymous, Toyota Seat Belt System Repair Manual for 1974 Models 3 Toyota 

Motor Sales Co., LTD., August 1973. 

12. Anonymous, Ford 1974 Shop Manual 3 Volume 4 - Body 3 Ford Marketing Corpora- 

tion, February 1974. 



i. 



j. 



* 

f- 



f 

ft 



I 




_ 



6.0 IMPLEMENTATION PLAN 



6.1 Introduction 

Three implementation plans for the evaluation of FMVSS 214, FMVSS 215, 
FMVSS 30.1 and FMVSS 208 are presented in this section. In developing the three 
plans, varying emphasis was placed on the following five considerations: 

(1) Schedule tasks which require existing data first. 

(2) Smooth out budget requirements, but spend more in first two 

years than in last two years. 

(3) Schedule field collection of new data last. 

(4) Obtain definitive conclusions on Standard effectiveness as 

soon as possible. 

(5) Consider probability of obtaining useable results in ordering 

tasks . 

Obviously, all of the above five considerations can not be satisfied simultan- 
eously. In the discussions that follow, the principal rationale and consider- 
ations that underlie each of the three plans are given. The three implementa- 
tion plans that are compared in this section are the following: 

• Early Results, Non- Integrated Plan 

• Integrated, Reduced Cost Plan 

• Early Results and Equalized Funding Plan. 

The total resources required to evaluate the four Standards are given in 
Table 6-1. The three categories of resource requirements are personnel, data 
processing and other costs such as data collection, personnel training and sur- 
vey mailings. An overall dollar cost is obtained by assuming $50,000 would be 
required for each person-year needed on a task. The overall costs for evalu- 
ating FMVSS 301 and FMVSS 208 are considerably higher than the estimated costs 
of the FMVSS 214 and FMVSS 215 evaluations. A significant portion of the total 
cost of evaluating FMVSS 301 and FMVSS 208 (about $600,000 needed for each 
Standard) is due to requirements for collecting new data. Three tasks, each 
estimated to require about $250,000, involve extensive data collection: analy- 

sis of fuel system rupture (FMVSS 301); analysis of passive system effective- 
ness (FMVSS 208); and analysis of seat belt usage (FMVSS 208). 

The more limited requirements for new data collection are largely respon- 
sible for keeping the estimated costs for evaluating FMVSS 214 under $500,000, 
and the estimated costs for evaluation of FMVSS 215 are less than $350,000 for 
the same reason. 

The total resources needed to evaluate the four Standards are estimated 
to be slightly in excess of two million dollars. This estimate is obtained 
from personnel requirements of 35.1 person-years (at $50,000 per person-year), 
$49,000 for data processing and $204,000 for other costs, mainly resulting from 
the data collection and acquisition efforts. 



6-1 



TABLE 6-1 

RESOURCES REQUIRED FOR EVALUATION OF STANDARDS 



Federal 

Motor 






Resources Required 


Total 


Vehicle 




Task 


Person- 


Data 


Other 


Cost 


Safety 






Years 


Processing 


Costs 




Standard 








($000) 


($000) 


($000) 




FMVSS 214: 


1 . 


Mass Accident Data 


1.0 


5 




55 




Analysis 










Side 


2. 


NCSS Data Analysis 


2.0 


8 




108 


Door 

Strength 


3. 


Field Accident (towaway) 


5.0 


5 


10 


265 






(NCSS type) 












4. 


Hardware Cost Analysis 


1.0 


1 




51 








9.0 


19 


10 


479 


FMVSS 215: 


1 . 


State Farm Insurance 
Data Analysis 


0.5 


1 




26 


Exterior 


2. 


Mass Accident Data 


0.5 


3 




28 


Protection 


3. 


Analysis 

HLDI Data Analysis 


0.5 


1 




26 




4. 


Car Owner Survey 


1.6 


3 


65 


148 




5. 


Towaway Survey 


0.5 


1 


30 


56 




6. 


Hardware Cost Analysis 


1.0 


1 




51 








4.6 


10 


95 


335 




1 . 


Fuel System Rupture 


4.5 


2 


13 


240 


FMVSS 301: 


2. 


(towaway) 

Fire/Fuel Spillage Analysis 


2.5 


3 


10 


138 


Fuel 




(Fi re Dept. ) 










System 


3. 


Fire-Related Fatalities 


3.0 


4 


10 


164 


Integrity 




(State & FARS Fatal 
Accident Data) 












4. 


Hardware Cost Analysis 


1.0 


1 




51 








11.0 


10 


33 


593 




1 . 


Seat Belt Effectiveness 


1.0 


2 


1 


53 


FMVSS 208: 




Analysis (RSEP/NCSS) 


4.0 




50 


255 




2. 


Passive System Effectiveness 


5 


Occupant 




Analysis (Existing & 










Crash 




Future Data) 










Protection 


3. 


Seat Belt Usage Survey 


4.5 


2 


15 


242 




4. 


Hardware Cost Analysis 


1.0 


1 




51 








10.5 


10 


66 


601 






Total Cost 


35.1 


49 


204 


2008 



6-2 



The annual funding required throughout a four year period for each of the 
three evaluation plans is shown graphically in Figure 6-1. It should be noted 
that the total funding required for the Integrated, Reduced Cost Plan is 
$1,725,000, a reduction of 14 percent from $2,008,000. This cost savings is 
achievable through proper time-sequencing and grouping of like tasks or tasks 
which require a common data base. 

The distribution of funding over the four year period of the evaluation 
project is significantly different among the three plans. In the Early Results, 
Non- Integrated Plan nearly all tasks are begun simultaneously at the start of 
the study. This plan emphasizes the desirability of obtaining definitive con- 
clusions on Standard effectiveness as soon as possible. In fact, final results 
for the evaluation of FMVSS 215 and FMVSS 301 are available by the first half 
of the second year and all but two tasks in FMVSS 214 and FMVSS 208 are com- 
pleted within the same time period. However, this plan has at least two very 
questionable characteristics. The non-sequential scheduling of almost all 
tasks will not allow much interactive use of results and analyses among tasks. 
Furthermore, the real-world budget and personnel constraints may not permit 
the allocation of over $1,400,000 to the first year of the project with a sub- 
sequent drastic reduction in funding levels. 

The Integrated, Reduced Cost Plan emphasizes different priorities in sched- 
uling tasks. The majority of tasks- scheduled during the first year make use 
only of existing data. Tasks requiring new field data collections are gener- 
ally begun in the second or third year of the project. The sequencing of tasks 
considers, where possible, the estimated probability of obtaining useable re- 
sults. As Figure 6-1 shows, the funding requirements are about $600,000 during 
each of the first two years, and much less during the final two years of the 
project. However, it must be noted that this plan has one potential serious 
drawback. The final definitive results on the evaluation of each of the four 
Standards will not be available until the fourth year of the project. This 
characteristic may not be acceptable when, for example, NHTSA considers how the 
results will be used in relation to other projects currently underway or planned. 

The final evaluation plan presented, Early Results and Equalized Funding 
Plan, is an attempt to retain the more desirable features of the first two 
plans, while at the same time eliminating their major differences. In this 
plan, the objective is to obtain relatively early results and to equalize the 
funding over the first three years of the project, with a drastic reduction in 
funding in the fourth year. To achieve these dual objectives, the work is time- 
sequenced according to Standard. All tasks under FMVSS 214 and FMVSS 215 are 
completed within the first two years. The FMVSS 301 effort will be conducted 
during the second and third years of the project and the work for FMVSS 208 will 
be undertaken during the third and fourth year. With this schedule, final re- 
sults on two of the Standards are available within the first two years of the 
evaluation project. The funding required is slightly in excess of $600,000 
in each of the first three years. While logical time-sequencing of tasks within 
each Standard will be retained, many of the cost saving features of the second 
plan may not be realized in the Early Results and Equalized Funding Plan, due 
to the staggering of the work schedule by Standards. 



6-3 



© Early Results, Non-Integrated Plan 




© Integrated, Reduced Cost Plan 




• Early Results and Equalized Funding Plan 




Year of Project 



Figure 6-1. Annual funding required for three evaluation plans. 



6-4 






Table 6-2 presents the final completion date in months after project start 
for each Standard in each of the three evaluation plans. Considering the criter- 
ion of obtaining early definitive final results, the Early Results and Equal- 
ized Funding Plan is comparable to the Early Results, Non-Integrated Plan. 

FMVSS 215 and FMVSS 208 are completed at the same time in both plans. FMVSS 
214 is completed twelve months earlier in the Early Results and Equalized Fund- 
ing Plan, while FMVSS 301 is completed fifteen months later in this plan. This 
parity in timely conclusion of Standards’ evaluation is achieved in the Early 
Results and Equalized Funding Plan without the highly skewed funding distribu- 
tion that occurs in the Early Results, Non-Integrated Plan. 

TABLE 6-2 

COMPLETION DATES FOR STANDARDS 



Federal 

Motor 

Vehicle 

Safety 

Standards 


Completion After Project Go-Ahead 


Early Results 
Non-Integrated 
(months) 


Integrated, 
Reduced Cost 
(months) 


Early Results and 
Equalized Funding 
(months) 


FMVSS 214: 

Side 

Door 

Strength 


36 




45 


24 


FMVSS 215: 

Exterior 
Protecti on 


16 


40 


16 


FMVSS 301: 

Fuel 
System 
Integri ty 


18 


42 


33 


FMVSS 208: 

Occupant 

Crash 

Protection 


48 


48 


48 



6.2 Early Results , Non-Integrated Plan 



The schedule and costs of the Early Results, Non-Integrated Plan for each 
of the four Standards to be evaluated is given in Figure 6-2, A total of 14 
tasks is included to evaluate the effectiveness of the four Standards and four 
tasks are required to determine the hardware costs of each Standard. As the 
title of this plan indicates, the tasks within and among the Standards are, 
for the most part, neither integrated nor time-sequenced. With the exception 
of the NCSS data analysis to determine side door strength (FMVSS 214) and the 
evaluation of passive system effectiveness (FMVSS 208), all tasks begin simul- 
taneously at the beginning of the study. Thus, 70 percent of the total project 
cost of $2,008,000 is concentrated in the first year of the study ($1,404,000). 
This very intensive effort during the first year of the study does produce the 
final results for the evaluation of FMVSS 215 (Exterior Protection) and FMVSS 
301 (Fuel System Integrity) within the first half of the second year, as well 
as the completion of all but two tasks in FMVSS 214 (Side Door Strength) and 
FMVSS 208 (Occupant Crash Protection) within the same time period. However, 
this essentially non-sequential scheduling of tasks does not permit much inter- 
active use of results and analyses among tasks. Further, real-world budgeting 
constraints may not permit such a highly skewed application of funding to the 
project. Work in the last two years of the project (50 percent of the time) 
requires only about 10 percent of the total resources. 



6-6 



Federal 

Motor 

Vehicle 

Safety 

Standard 


Task 


Data 

Availability 


Task 

Duration 


Time After Project Go-Ahead (Years) 


Cost 

($000) 


Old 

Data 


New 

Data 


(Months) 


Year 
1 L 


1 


Year 2 

1 I I 


Year 3 

l i I 


Year 4 

_ l i l 


r 


















a 
















55 


FMVSS 214: 


1. 


Mass Accident Data 
Analysis 


X 




y///A 


























L 


I 
















6ide 


2. 


NCSS Data Analysis 


X 




9 




m mm 












108 


Door 




Field Accident (towaway) 
(NCSS type) 






19 






1 












265 


Strength 


3. 




X 


















X 




















51 




4. 


Hardware Cost Analysis 




7 


wm 


\ 










































479 




1. 


State Farm Insurance 
Data Analysis 
























26 


FMVSS 215: 


X 




6 


WA 










































28 i 


Exterior 


2. 


Mass Accident Data 
Analysis 


X 




6 


W/A 
















Protection 




X 


























3. 


HLDI Data Analysis 




6 


















26 










X 


12 


















148 




4. 


Car Owner Survey 




V/////////A 


















Towaway Survey 




X 


















56 , 




5. 






1 0 


















Hardware Cost Analysis 




X 


7 
















51 




0. 








S/S/Ss 


2 














































" 










| 














335 












18 














240 




1. 


Fuel System Rupture 
(towaway) 


(X) 


X 


msmsmm 












FMVSS 301: 
























138 




2. 


Fire/Fuel Spillage Analysis 


X 




12 


W////M 














Fuel 




(Fire Dept.) 






12 


1 














164 


System 


3. 


Fire-Related Fatalities 
(State & FARS Fatal 
Accident Data) 


X 




WMM, 
























1 


















4. 


Hardware Cost Analysis 




X 


7 


















51 


























593 




1. 


Seat Belt Effectiveness 
Analysis (RSEP/NCSS) 


X 




9 


mm 














53 


FMVSS 208: 


* 
























255 




2. 


Passive System Effectiveness 


X 


X 


25 






V///A 


W////A 


Occupant 

Crash 




Analysis (Existing & 
Future Data) 










i 














Protection 


3. 


Seat Belt Usage Survey 




X 


15 


\y//////y/////yM 












242 








\r 




• 














51 




4. 


Hardware Cost Analysis 




X 


7 




m 










































601 




Total Cost ($000) 


1,404 


391 


166 


47 


2008 



Figure 6-2. Schedule and costs of Early Results, Non-Integrated Plan. 



6-7 









6,3 Integrated, Reduced Cost Plan 



The schedule and costs of the Integrated, Reduced Cost Plan for the four 
Standards are given in Figure 6-3. The premises for formulating the plan and 
the resultant schedule are quite different from the first plan presented. Al- 
though work is conducted simultaneously under all four Standards, the majority 
of tasks scheduled during the first year require only existing data. Those 
tasks which require new data collection or extensive data acquisition are gen- 
erally scheduled to start in thr* second or third year of the project. One con- 
sideration taken into account (when possible) in the scheduling of tasks is the 
estimated probability of obtaining useful results. Other factors, however, may 
override this consideration. For example, the analysis of fuel system rupture, 
the most expensive task in the evaluation of FMVSS 301, and also the task judged 
most likely to produce useful results, is not scheduled to begin until the 
third year, as all towaway data collection tasks are scheduled in common during 
the third and fourth year of the project. 

The cost reductions of $283,000 or 14 percent that are achieved in the 
Integrated, Reduced Cost Plan are due to simultaneous scheduling of tasks to 
be undertaken by a single agency or organization. These tasks which depend on 
a common data base, require a similar analysis methodology, or involve related 
new data collection efforts are: (1) analyses requiring mass accident data, 
State Farm data and HLDI data; (2) the hardware cost analysis for each Standard; 
(3) analyses utilizing NCSS and RSEP data; (4) data collection efforts involving 
towaway accidents, and (5) analyses of fire/fuel spillage and fire-related fa- 
talities . * 

The funding requirements for the Integrated, Reduced Cost Plan are close 
to $600,000 during each of the first two years and drop to about $350,000 and 
$175,000 respectively during the last two years of the project. Thus, this plan 
achieves both cost reductions and a steady level of funding during the first 
two years which is reduced during the third and fourth year of the project. 

The Integrated, Reduced Cost Plan does contain at least one potentially 
serious drawback. The final definitive results of the evaluation of each of 
the four Standards will not be available until the fourth year of the project. 

It is true, of course, that substantial and perhaps rather conclusive inter- 
mediate results will be available well before the end of the fourth year. How- 
ever, this mode of planning may not be acceptable to NHTSA if final definitive 
results are needed sooner because of the demands and requirements of other pro- 
jects currently underway or planned. 



The analyses of fire/fuel spillage and fire-related fatalities are scheduled 
to be undertaken sequentially rather than simultaneously. However, cost re- 
ductions can be realized if these tasks are conducted by the same agency or 
organization. 



6-8 




Dollar amount in parenthesis is the expected cost associated with the Non-Integrated, Early Results Plan. It 
is provided to permit easy comparison of the two plans. 

Figure 6-3. Schedule and costs of Integrated, Reduced Cost Plan. 

■ 






6-9 



6.4 Early Results and Equalized Funding Plan 



The schedule and costs of the Early Results and Equalized Funding Plan 
are given in Figure 6-4. The scheduling in this plan was formulated in an 
attempt to retain the more desirable features of the first two plans discussed 
while at the same time eliminating their major deficiencies. In this plan 
the basic objective is to obtain relatively early definitive final results 
for some Standards and to equalize the funding over the first three years of 
the project. This plan requires a funding level slightly in excess of 
$600,000 in each of the first three years. The resource requirements drop 
drastically in the fourth year to about $125,000. 

The above results are achieved by (1) scheduling work in a time- 
sequenced framework according to Standard and (2) modifying (reducing) where 
this is feasible and helpful with regard to the length of time required to 
accomplish the evaluation work under a given Standard. Only tasks under FMVSS 
215 and FMVSS 214 are initiated during the first years. All work under these 
Standards is completed within the second year. Thus, the definitive final 
results of two of the four Standards being evaluated are scheduled to be 
available within the first two years of the project. to achieve this result, 
the total time slapsed for evaluating FMVSS 214 has been reduced from three 
years to two years. This reduction appears to be entirely feasible by (1) 
beginning the NCSS data acquisition and analysis at the beginning of the study 
rather than waiting six months for the completion of the mass accident data 
analysis; (2) beginning the Field Accident (NCSS-type) data collection seven 
months after the start of the study; and (3) allowing 18 months (rather than 
21 months) for data collection, preparation, analysis and reporting. 

The FMVSS 301 effort will be conducted during the second and third year 
of the project. The work under FMVSS 208 will not begin until the third year 
and will be completed by the end of the fourth year. One could justify the 
delay in evaluating because (1) much more accident data on passive restraint 
systems will be available and (2) the study of active restraint system usage 
will reflect usage patterns that are representative of the very late 1970’s 
with a greater preponderance of 3-point lap/shounder belts in the car popula- 
tion. The elapsed time for conducting the evaluation of FMVSS 208 is reduced 
from four years to two years in the EArly Results and Equalized Funding Plan. 
This can be accomplished by (1) beginning the analysis of passive system effec- 
tiveness at the start of the study (9 months time saved) and (2) eliminating 
one sequence of a more comprehensive re-analysis of the passive system accident 
data (two re-analyses are included in the original plan) and/or eliminating 
the two 6-month gaps of inactivity between re-analyses (12-18 months time 
saved) . 

Due to the time sequencing of work by Standard, most of the cost saving 
features of the Integrated, Reduced Cost Plan cannot be realized in this third 
plan. On the other hand, the Early Results and Equalized Funding Plan does 
have the desirable characteristic of providing definitive final results on two 
Standards within two years of the start of the project, and also maintains a 
steady level of funding during the first three years, when most of the work is 
accomplished . 



6-10 



Federal 

Motor 

Vehicle 

Safety 

Standard 


Time After Project Go-Ahead (Years) 


Cost 

($000) 


Year 1 Year 2 

1 ■ • J 1 _ ..I 1 1 


Year 3 

1 1 1 


Year 4 

1 1 1 


FMVSS 215: 

Exterior 

Protection 








335 




n 


.1.1 


.in. nnnn 




FMVSS 214: 

Side 

Door 

Strength 








479 


n 


H 




i 




- 




FMVSS 301: 

Fuel 

System 

Integrity 


I 






593 


FMVSS 208: 

Occupant 

Crash 

Protection 








601 


I 


I"" 






Total Cost 
($000) 


608 619 


657 


124 


2008 



Figure 6-4. Schedule and costs of Early Results and Equalized Funding Plan. 



6-11 



7 . 0 END PRODUCTS OF THIS STUDY 



During this study, Evaluation Methodology for Four Federal Motor Vehicle Safety 
Standards (Contract DOT-HS-6-01518) , ten reports and two briefings were pre- 
pared between October 1976 and March 1977. In addition to those materials 
(listed below) , many special appendices were assembled. 

• CEM Report 4207-559. Review of Four Federal Motor Vehicle Safety 

Standards: FMVSS 214 s 215, 301 , 208 , October 1976. 

• CEM Report 4207-560. Preliminary Design on an Evaluation Procedure 

for FMVSS 214: Side Door Strength , November 1976. 

• CEM Report 4207-561. Preliminary Design of an Evaluation Procedure 

for FMVSS 215: Exterior Protection , November 1976. 

• CEM Report 4207-562. Preliminary Design of an Evaluation Procedure 

for FMVSS 301: Fuel System Integrity , December 1976. 

• CEM Report 4207-563. Preliminary Design of an Evaluation Procedure 

for FMVSS 208: Occupant Crash Protection , December 1976. 

These preliminary reports contained copies of the latest version of the Standard 
an appendix describing the cost estimating methodologies of the Bureau of Labor 
Statistics, Government Accounting Office, and NHTSA; and specific appendices 
on temporary exemption from Standards, introduction dates of side door reinforce 
ment beams, and a statistical discussion about selecting make and model for data 
sampling . 

• CEM Report 4207-564. Final Design and Implementation Plan for Evalu- 

ating the Effectiveness of FMVSS 214: Side Door Strength , 

January 1977. 

• CEM Report 4207-565. Final Design and Implementation Plan for Evalu- 

ating the Effectiveness of FMVSS 215: Exterior Protection , 

February 1977. 

e CEM Report 4207-566. Final Design and Implementation Plan for Evalu- 
ating the Effectiveness of FMVSS 301: Fuel System Integrity, 

February 1977. 

• CEM Report 4207-567. Final Design and Implementation Plan for Evalu- 

ating the Effectiveness of FMVSS 208: Occupant Crash Protection, 

March 1977. 

These detailed reports contain general appendices which discuss several statis- 
tical methods, and the proposed Standard implementation cost categories-r Also, 
many specific appendices are on the anticipated distribution of AIS levels 
in sampled accident data, rate of return for surveys, distribution of pre-and 
post- FMVSS 301 vehicles in fatal accidents by accident year, and discussion of 
contingency table analysis. 



7-1 



The Final Report and briefings are: 



• CEM Report 4207-568. Evaluation Methodologies for Four Federal Motor 

Vehicle Safety Standards: FMVSS 214 - Side Boor Strength; FMVSS 
215 - Exterior Protection; FMVSS 301 - Fuel System Integrity; 
FMVSS 208 - Occupant Crash Protection , March 1977. 

« CEM DWN 887. Interim Report on Final Design and Implementation Plans 
for Evaluation of Federal Motor Vehicle Safety Standards: 

FMVSS 214 Side Door Strengths FMVSS 215 - Exterior Protection 
(Briefing), 16 February 1977. 

• CEM DWN 892, Final Report on Final Design and Implementation Plans 

for Evaluation of Federal Motor Vehicle Safety Standards: 

FMVSS 301 - Fuel System Integrity; FMVSS 208 - Occupant Crash 
Protection (Briefing), 31 March 1977. 



7-2 



APPENDIX A 

FEDERAL MOTOR VEHICLE 
SAFETY STANDARD NO, 214 



Effective: January 1, 1973 



APPENDIX A. MOTOR VEHICLE SAFETY STANDARD NO. 214 



MOTOR VEHICLE SAFETY STANDARD NO. 214 

Side Door Strength-— Passenger Cars 
(Docket No. 2—6; Notice No. 3) 



SI. Purpose and scope. This standard speci- 
fies strength requirements for side doors of a 
motor vehicle to minimize the safety hazard 
caused by intrusion into the passenger compart- 
ment in a side impact accident. 

$2. Application. This standard applies to pas- 
senger cars. 

53. Requirements. Each vehicle shall be able 
to meet the following requirements when any of 
its side doors that can be used for occupant egress 
are tested according to S4. 

53.1 Initial crush resistance. The initial crush 
resistance shall be not less than 2,250 pounds. 

53.2 Intermediate crush resistance. The inter- 
mediate crush resistance shall not be less than 
3,500 pounds. 

$3.3 Peak crush resistance. The peak crush 
resistance shall be not less than two times the 
curb weight of the vehicle or 7,000 pounds, 
whichever is less. 

54. Test procedures. The following procedures 
apply to determining compliance with section 
S3: 

(a) Remove from the vehicle any seats that 
may affect load upon, or deflection of, the side of 

he vehicle. Place side windows in their upper- 
most position and all doors in locked position. 
Place the sill of the side of the vehicle opposite 
to the side being tested against a rigid unyield- 
ing vertical surface. Fix the vehicle rigidly in 
position by means of tiedown attachments lo- 
cated at or forward of the front wheel center- 
line and at or rearward of the rear wheel center- 
line. 

(b) Prepare a loading device consisting of a 
rigid steel cylinder or semi -cylinder 12 inches in 
diameter with an edge radius of one-half inch. 



The length of the loading device shall be such 
that the top surface of the loading device is at 
least one-half inch above the bottom edge of the 
door window opening but not of a length that 
will cause contact with any structure above the 
bottom edge of the door window opening during 
the test. 

(c) Locate the loading device as shown in 
Figure I (side view) of this section so that: 

(1) Its longitudinal axis is vertical; 

(2) Its longitudinal axis is laterally op- 
posite the midpoint of a horizontal line drawn 



CENTER LINE OF VEHICLE 




SIDE VIEW 



LOADING DEVICE LOCATION AND APPLICATION TO THE DOOR 

FICUSS l 

across the outer surface of the door 5 inches 
above the lowest point of the door; 

(3) Its bottom surface is in the same hori- 
zontal plane as the horizontal line described 
in subdivision (2) of this subparagraph; and 



PART 571; S 214-1 





Effective* January t, 1973 



(4) The cylindrical face of the device is in 

contact with the outer surfaco of the door. 

(d) Using the loading device, apply a load to 
the outer surface of the door in an inboard di- 
rection norma) to a vertical plane along the 
vehicle's longitudinal centerline. Apply the 
load continuously such that the loading device 
travel rate does not exceed one-half inch per 
second until the loading device travels 18 inches. 
Guide the loading device to prevent it from 
being rotated or displaced from its direction of 
travel. The test must be completed within 120 
seconds. 

(e) Record applied load versus displacement 
of the loading device, either continuously or in 
increments of not more than 1 inch or 200 pounds 
for the entire crush distance of 18 inches. 

(f) Determine the initial crush resistance, in- 
termediate crush resistance, and peak crush re- 
sistance as follows : 



(1) From the results recorded in subpara- 
graph (o) of this paragraph, plot a curve of 
load versus displacement and obtain the in- 
tegral of the applied load with respect to the 
crush distances specified in subdivisions (2) 
and (3) of this paragraph. These quantities, 
expressed in inch-pounds and divided by the 
specified crush distances, represent the average 
forces in pounds required to deflect the door 
those distances. 

(2) The initial crush resistance is the aver- 
age forco required to deform the door over the 
initial 6 inches of crush. 

(3) The intermediate crush resistance is the 
average force required to deform the door over 
the initial 12 inches of crush. 

(4) The peak crush resistance is the largest 
force recorded over the entire 18-inch crush 
distance. 

October 30, 1970 
35 F.R. 16801 



PART 571; S 214-2 
A- 3 



APPENDIX B 

FEDERAL MOTOR VEHICLE 
SAFETY STANDARD NO. 215 



Effective: September 1, 1972 
September 1, 1973 



APPENDIX A: MOTOR VEHICLE SAFETY STANDARD NO. 215 



MOTOR VEHICLE SAFETY STANDARD NO. 215 
Exterior Protection — Passenger Cars 
(Docket Nos. 1-9 and 1—10; Notice No. 4) 



51. Scope. This standard establishes require- 
ments for the impact resistance and the con- 
figuration of front and rear vehicle surfaces. 

52. Purpose. The purpose of this standard is 
to prevent low-speed collisions from impairing 
the safe operation of vehicle systems, and to re- 
duce the frequency of override or underride in 
higher speed collisions. 

53. Application. This standard applies to pas- 
senger cars. 

54. Definition. All terms defined in the Act 
and the ndes and standards issued under its au- 
thority are used as defined therein. 

55. Requirements. 

55.1 Vehicles manufactured on or after Septem- 
ber 1, 1972. 

Each vehicle manufactured on or after Sep- 
tember 1, 1972, shall meet the protective criteria 
of S5.3.1 through S5.3.4 when it impacts a fixed 
collision barrier that is perpendicular to the line 
of travel of the vehicle, while traveling longitu- 
dinally forward at 5 mph and while traveling 
longitudinally rearward at 2y 2 mph, under the 
conditions of S6.1. 

55.2 Vehicles manufactured on or after Sep- 
tember 1, 1973. 

[Except as provided in S5.2.1 and So.2.2, each 
vehicle manufactured on or after September 1, 
1973, shall meet the protective criteria of S5.3.1 
through So. 3.7 during and after impacts by a 
pendulum- type test device in accordance with the 
procedures of S7.1 and S7.2 followed by impacts 
into a fixed collision barrier that is perpendicular 
to the line of travel of the vehicle, while travel- 
ing longitudinally forward at 5 mph and while 
traveling longitudinally rearward at 5 mph un- 
der the conditions of S6. (39 F.R. 29369 — 

August 15, 1974. Effective: 3/1/75)] 

(Rev. 5/7/75) 



55.2.1 [The corner-impact procedure of S7.2.2 
shall not apply to any vehicle with a wheelbase 
exceeding 120 inches manufactured from Septem- 
ber 1. 1973 to August 31. 1976. (40 F.R. 20823- 
May 13, 1975. Effective: 5/13/75)] 

55.2.2 [The fixed collision barrier impact re- 
quirements of S5.2 shall apply, but the pendulum 
impact requirements of S5.2 shall not apply to 
each vehicle manufactured from September 1, 
1973 to October 31. 1974, that has a wheelbase 
of 115 inches or less and that either — 

(a) Has a convertible top; 

(b) Has no roof support structure between the 
A-pillar and the rear roof support structure: or 

(c) Has no designated seating position behind 
the front designated seating positions. 

(39 F.R. 31641 — August 30, 1974. Effective: 
9/1/74)] 

S5.3 Protective criteria. 

55.3.1 [Each lamp or reflective device, except 
license plate lamps, shall be free of cracks and 
shall comply with the applicable visibility re- 
quirements of S4.3.1.1 of Standard Xo. 108 
(§ 571.108 of this part). The aim of each head- 
lamp shall be adjustable to within the beam aim 
inspection limits specified in Table 2 of SAE 
Recommended Practice J599b, July 1970. meas- 
ured with a mechanical aimer conforming to the 
requirements of SAE Standard J602a, July 1970. 
(37 F.R. 16803 — August 19, 1972. Effective: 
9/1/72)] 

55.3.2 The vehicle's hood, trunk, and doors 
shall operate in the normal manner. 

55.3.3 T1 le vehicle's fuel and cooling systems 
shall have no leaks or constricted fluid passages 
and all sealing devices and caps shall operate in 
the normal manner. 



PART 571; S 215-1 



B-2 



(ffactiva: Sapt*mb«r 1, 1972 
September 1, 1973 



55.3.4 The vehicle's exhaust system shall have 
no leaks or constrictions. 

55.3.5 [The vehicle's propulsion, suspension, 
steering, and braking systems shall remain in ad- 
justment and shall operate in the normal manner. 
(36 F.R. 23802 — December 15, 1971. Effective: 
September 1, 1972)]* 

55.3.6 The vehicle shall not touch the test de- 

vice except on the impact ridge shown in Figures 
1 and 2. (36 F.R. 20369— October 21, 1971. 

Effective: 9/1/72, except as noted in S5.2)] 

[55.3.7 A pressure vessel used to absorb im- 
pact energy in an exterior protection system by 
the accumulation of gas pressure or hydraulic 
pressure shall not suffer loss of gas or fluid ac- 
companied by separation of fragments from the 
vessel. (39 F.R. 29369 — August 15, 1974. Ef- 
fective: 3/1/75)3 

S6. Conditions. The vehicle shall meet the 
requirements of S5 under the following condi- 
tions. 

56.1 General. 

56.1.1 The vehicle is at unloaded vehicle 
weight. 

56.1.2 The front wheels are parallel to the 
vehicle’s longitudinal centerline. 

56.1.3 Tires are inflated to the vehicle manu- 
facturer’s recommended pressure for the specified 
loading condition. 

56.1.4 Brakes are disengaged and the trans- 
mission is in neutral. 

[S6.1.5 Trailer hitches are removed from the 
vehicle. (37 F.R. 16803 — August 19, 1972. Ef- 
fective: 9/1/72)3 

S6.2 Pendulum test conditions. The follow- 
ing conditions apply to the pendulum test pro- 
cedures of S7.1 and S7.2. 

56.2.1 The test device consists of a block with 
one side contoured as specified in Figure 1 and 
Figure 2 with the impact ridge made of hardened 
steel. 

56.2.2 With plane A vertical, the impact line 
shown in Figures 1 and 2 is horizontal at the 
same height as the test device’s center of percus- 
sion. 

•S5.2 through S5.3.6 were amended October 21, 1971. 
S5.3.1 and S5.3.5 were subsequently amended 36 F.R. 
23802— December 15, 1971 



TOP VIEW 





FIGURE 1 



56.2.3 The effective impacting mass of the test 
device is equal to the mass of the tested vehicle. 

56.2.4 When impacted by the test device, the 
vehicle is at rest on a level, rigid concrete surface. 

[S6.3 Barrier test condition. At the onset of a 
barrier impact, the vehicle’s engine is operating 
at idling speed. (36 F.R. 20369 — October 21, 
1971. Effective: 9/l/72)J 

57. Test procedures. 

S7.1 Longitudinal impact test procedures. [Im- 
pact the vehicle’s front surface and its rear sur- 
face two times each with the impact line at any 
height between 20 inches and 16 inches, in accord- 
ance with the following procedure. (40 F.R. 
20823— May 13, 1975. Effective date: 5/13/75)3 



TOP VIEW 





FIGURE 2 



OUv. 5/7/75) 



PART 571; S 215-2 



B-3 



Effective: September 1, 1972 
September 1, 1973 



57.1.1 For impacts at a height of 20 inches, 
place the test device shown in Figure 1 so that 
plane A is vertical and the impact line is hori- 
zontal at the specified height. 

57.1.2 For impacts at a height between 20 
inches and 16 inches, place the test device shown 
in figure 2 so that plane A is vertical and the im- 
pact line is horizontal at a height within the 
range. 

57.1.3 For each impact, position the test de- 
vice so that the impact line is at least 2 inches 
apart in vertical direction from its position in 
any prior impact, unless the midpoint of the 
impact line with respect to the vehicle is to be 
more than 12 inches apart laterally from its 
position in any prior impact. 

57.1.4 For each impact, align the vehicle so 
that it touches, but does not move, the test device, 
with the vehicle's longitudinal centerline per- 
pendicular to the plane that includes plane A of 
the test device and with the rest device inboard 
of the vehicle corner test positions specified in 
S7.2. 

ES7.T.5 Move the test device away from the ve- 
hicle, then release it so that plane A remains 
vertical from release until the onset of rebound, 
and the arc described by any point on the im- 
pact line is constant, with a radius of not less 
than 11 feet, and lies in a plane parallel to the 
vertical plane through the vehicle's longitudinal 
centerline. (36 F.R. 8734 — May 12, 1971)] 

57.1.6 [Impact the vehicle at 5 mph. (36 

F.R. 20369 — October 21, 1971. Effective: 

9/1/72)] 

57.1.7 Perform the impacts at intervals of not 
less than 30 minutes. 



S7.2 Corner impact test procedure. Impact a 
front corner and a rear corner of the vehicle 
once each with the impact line at a height of 20 
inches an 1 impact the other front corner and 
the other rear corner once each with the impact 
line at any height between 20 inches and 16 
inches in accordance with the following pro- 
cedure. 

57.2.1 For an impact at a height of 20 inches, 
place the test device shown in figure 1 so that 
plane A is vertical and the impact line is hori- 
zontal at the specified height. 

57.2.2 For an impact at a height between 20 
inches anil 16 inches, place the test device shown 
in figure 2 so that plane A is vertical and the 
impact line is horizontal at a height within the 
range. 

57.2.3 Align the vehicle so that a vehicle 
corner touches, but does not move, the lateral 
center of the test device with plane A of the 
test device forming an angle of 60 degrees with 
a vertical longitudinal plane. 

57.2.4 Move the test device away from the 
vehicle, then release it so that plane A remains 
vertical from release until the onset of rebound, 
and the arc described by any point on the im- 
pact line is constant, with a radius of not less 
than 11 feet, and lies in a vertical plane at an 
angle of 30° to the vertical plane through the 
vehicle's longitudinal centerline. 

57.2.5 Impact each corner at 3 mph. 

36 F.R. 7218 
April 16, 1971 

36 F.R. 8734 
May 12, 1971 



(S«v. Oct. 1971) 



PART 571; S 215-3 



B-4 



APPENDIX C 

FEDERAL MOTOR VEHICLE 
SAFETY STANDARD NO. 301 



RULES AND REGULATIONS 



p. 48352 



(Docket No. 73-20; Notice 8] 

PART 571— MOTOR VEHICLE SAFETY 
STANDARDS 

Fuel System Integrity 

The purpose of this notice Is to amend 
Motor Vehicle Safety Standard No. 301, 
Fuel System Integrity (49 CFR 571 .301) 
to extend the applicability of the stand- 



FEDERAl REGISTER, VOL 40, NO. 200 — WEDNESDAY, OCTOIEt 15, 1975 



C-2 




ard to school buses with a GVWR in 
excess of 10.000 pounds. The amendment 
specifies conditions for a moving con- 
toured barrier crash for school buses 
in order to determine the amount of 
fuel spillage following Impact. 

On October 27, 1974, the Motor Ve- 
! hide and Schoolbus Safety Amendments 
of 1974. amending the National Traffic 
and Motor Vehicle Safety Act. were 
signed into law (Pub. L. 93-492. 88 Stat. 
1470'. Section 103(1X1) (A) of the Act, 
as amended, orders the promulgation of 
a safety standard establishing minimum 
requirements for the fuel system integ- 
rity of school buses. Standard No. 301 
i currently contains requirements for 
school buses with a GVWR of 10.000 
! pounds or less which will become effec- 
tive beginning September 1. 1970. Larger 
school buses, which comprise approxi- 
mately 90 percent of the school bus pop- 
I ulation. will be Included in Standard No. 
301 by this amendment 

A proposal to amend Standard No. 301 
j with respect to school buses, loading con- 
ditions. and spillage measurement time 
was published on April 16. 1975 (40 FR 
I 17036). An amendment to the Standard 
I specifying certain loading conditions and 
i I establishing a 30-minute fuel spillage 
measurement period was published on 
August 6. 1975 (40 FR 33036). At the 
j request of several members of Congress, 
the period for comments on the school 
: bus proposals was extended This notice 
[ responds to the comments received with 
j! respect to the inclusion of school buses 
■within the requirements of the standard. 

Seven manufacturers opposed the re- 
1 1 quirement of a single impact test by a 
i moving contoured barrier at any point on 
lithe school bus bodv, arguing that such a 
I requirement would necessitate a prolif- 
{ eration of expensive tests in order to en- 
I I sure compliance at every conceivable 
point of impact The NHTSA does not 
I agree Although not specifying a partic- 
ular impact point, the test condition 
allows for testing at the few most vul- 
I nerable points of each kind X school bus 

I fuel system configuration Therefore. 
■I only impacts at those points are neces- 

jllsary to determine compliance On the 
basis of its knowledge of the bus design, 
a manufacturer should be able to make at 

I I least an approximate determination of 
' I the most vulnerable points on the bus 

body. 

Two school bus body manufacturers 
J' requested a requirement that the manu- 
facturer who installs the fuel svstem be 
responsible for compliance testing, while 
one chassis manufacturer argued that 
[ responsibility for compliance should rest 
with the final manufacturer In most 
cases, if the basic fuel system compo- 
nents are included in the chassis as de- 
livered by its manufacturer, the multi- 
stage vehicle regulations of 49 CFR Part 
568 require the chassis manufacturer at 
least to describe the conditions under 
which the completed vehicle will con- 
form, since it could not truthfullv state 
that the design of the chassis has no 
substantial determining effect on con- 
formity. Beyond that, however, the 
NHTSA position is that the decision as 
to who should perform the tests and who 



RULES AND REGULATIONS 

should take the responsibility is best not 
regulated by the government. The effect 
of Part 568 Is to allow the final-stage 
manufacturer to avoid primary responsi- 
bility for conformity to a standard if ft 
completes the vehicle in accordance with 
the conditions or instructions furnished 
with the incomplete vehicle by its manu- 
facturer. Whether it does so is a decision 
it must make in light of all the circum- 
stances. 

This notice extends the proposed ex- 
clusion for vehicles that use fuel with a 
boiling point below 32° F. to school buses 
having a GVWR greater than 10.100 
pounds. Fuel systems using gaseous fuels 
are not subject to the spillage problems 
against which this standard is directed. 

The Vehicle Equipment Safety Com- 
mission requested that school buses be 
required to undergo static rollover tests 
and that the engine be running during 
the tests. Upon consideration, the 
NHTSA finds that a static rollover test 
for schoolbuses is impractical in light of 
the expensive test facility that would be 
required. A requirement that the engine 
be running during the impact test would 
make little difference in the resulting 
fuel spillage Since the standard requires 
that the fuel tank be filled with Stoddard 
solvent during the impact test, the test 
vehicle would have to be equipped with 
an auxiliary fuel system lor the engine. 
The expense of modifying the test ve- 
hicle to allow the engine to run during 
the test would not justify the minimal 
benefits resulting from a reo.uirement 
that the engine be running How’ever, the 
fuel system integrity of school buses will 
be continually monitored and analyzed 
by the NHTSA. Therefore, suggestions 
such as these may be the subject of fu- 
ture rulemaking. 

One school bus body manufacturer 
cited the infrequency of school bus flies 
resulting from collisions as a reason for 
ameliorating or eliminating altogether 
fuel system integrity requirements for 
school buses. In promulgating these 
amendments to Standard No 301. the 
NHTSA is acting under the statutory 
mandate to develop regulations concern- 
ing school bus fuel svstems. This statute 
reflects the need, evidently strongly felt 
by the public, to protect the children who 
ride in the school buses They and their 
parents have little direct control over 
the types of vehicles in which they ride 
to school, and are therefore not in a por- 
tion to determine the safety of the ve- 
hicles. Considering the high regard ex- 
pressed by the public for the safety of 
its children, the NHTSA finds it Impor- 
tant that the schoolbus standards be 
effective and meaningful. 

The California Highway Patrol ex- 
pressed the concern that these amend- 
ments would preempt State regulations 
to the extent that the State would be pre- 
cluded from specifying the location of 
fuel tanks, fillers, vents, and drain open- 
ings in school buses. The standard will 
unavoidably have that effect, by the op- 
eration of section 103(d) of the National 
Traffic and Motor Vehicle Safety Act. 
However, although a State may not have 
regulations of general applicability that 



48353 

bear on these apsects of performance, 
the second sentence of the same section 
makes it clear that a State or political 
subdivision may specify higher standards 
of performance for vehicles purchased 
for its own use, although of course the 
Federal standards must be met in any 
case. 

In addition to provisions directly re- 
lating to schoolbuses. this notice clarifies 
the loading condition amendments in 
the notice of August 6, 1975, by amend- 
ing S5.1 to provide for testing with 50th 
percentile dummies. The wording of S6.1 
Is identical to that of the proposal. 

In light of the foregoing, 49 CFR 
571.301, Motor Vehicle Safety Standards 
No. 301, is revised to read as set forth 
below. 

Effective date: July 15. 1976, in con- 
formity with the schedule mandated by 
the 1974 Amendments to the Traffic 
Safety Act. However, the effective date 
of the amendment of S8.1 is October 15. 
1975. Because the amendment to that 
paragraph clarifies the revision of cer- 
tain requirements which became effec- 
tive September 1, 1975, it is found for 
good cause shown that an effective date 
for the amendment of S6.1 less than 
180 days after issuance is in the public 
interest. 

( Sec 103. 119, Pub L. 89-563. 80 Stat. 718 
(15 u s e. 1392, 1407): aec. 202. Pub. L. 93- 
492. 88 Stat. 1470 (15 U.S.C. 1392): delega- 
tions of authority at 49 CFR 1.51 and 601.8). 

Issued on October 8, lSTS. 

Gene G. Mannella, 
Acting Administrator. 

Section 571.301 is revised as follows: 

§ 571.301 Standard No. 301 ( fuel sys- 
tem integrity. 

51. Scope. This standard specifies re- 
quirements for the integrity of motor ve- 
hicle fuel systems. 

52. Purpose. The purpose of this stand- 
ard is to reduce deaths and injuries oc- 
curring from fires that result from fuel 
spillage during and after motor vehicle 
crashes. 

53. Application This standard applies 
to passenger cars, and to multipurpose 
passenger vehicles, trucks, and buses 
that have a GVWR of 10,000 pounds or 
less and use fuel with a boiling point 
above 32° F, and to schoolbuses that have 
a GVWR greater than 10,000 pounds and 
use fuel with a boilihg point about 32° F. 

54. Definition. '‘Fuel spillage'' means 
the fall, flow, or run of fuel from the 
vehicle but does not include wetness re- 
sulting from capillary action. 

55. General requirements. 

55.1 Passenger cars. Each passenger 
car manufactured from September 1, 
1975, to August 31, 1976, shall meet the 
requirements of S6.1 in a perpendicular 
impact only, and S6.4. Each passenger 
car manufactured on or after September 
1, 1976, shall meet all the requirements 
of S6, except S6.5. 

55.2 Vehicles with GVWR of 6.000 
pounds or less. Each multipurpose pas- 
senger vehicle, truck, and bus with a 
GVWR of 6.000 pounds or less manu- 
factured from September 1, 1976, to Au- 



FEDHRAL REGISTER, VOL 40, NO. 200 — WEDNESDAY. OCTOBER )5, 1975 

C-3 



•i Sl>5 1 



RULES AND REGULATION 



gust 31, 1977. shall meet all the require- 
ments of S6. . in a perpendicular impor: 
only, S6.2, and S6 4. Each of these types 
of vehicles manufactured on or after 
September 1 1977. shall meet all the re- 
quirements of SS. except S6 5. 

S5.3 Vehicles with GVWR o' more 
then 6.000 pounds hut not more than 

10.000 pounds Each multipin nose pas- 
senger vehicle, truck, and bus with a 
GVtVR of more than 6.000 pounds but 
not. more than. 10.000 pounds manufac- 
tured from September 1. 1976. to Au- 
gust 31, 1977. shall meet the require- 
ments of S6 l in a perpendicular impact 
only. Each vehicle manufactured on or 
after September 1. 1977. shall meet all 
the requirements of S6. except SO 5. 

S5 4 Schonlhvses with a GVWR greater 
then 10.000 pounds. Each schoolbuc with 
a GVWR creator than lO.oao pounds 
manufactured on or after July 15, 1970. 
shall meet the requirement? of SC 5 

So 5 Fuel spillnae: Barrier crash Fuel 
spilla--.- in anv fixed or moving barrier 
crash test shall not exceed 1 on nee bv 
weight from impact until motion of the 
vehicle has ceased and sh ill not exceed 
a tola) of 5 ounces bv weuht in t! e 
5-minute period following cessation of 
motion lor the snb c equcnt 25-mim:'' 
period fuel si illnge dtirit.a on; l-mh;titv 
ji.tervt! ;;h. i;c.‘ c\ ) ov.rce ! •• 

\« . out 

S3 C Fu-d sn.Vrifie. Rn Fuel ... . - 
latte in. any rollover test from the ••••■•■ : 
of rota‘iur;,l mo'i'.n, shall not cxrr, . 
a total of * otn-'-t ly wekiht for i he 
fir -l 3 minuter, cf tori ir.g at each site. 

i . e 90 increment For the rt mai; - 
im. testing period at each increment d 
90 fuel spillage during any 1-mlniitc 
interval shall not exceed 1 ounce l y 
weight 

3(5. Test requn en:cnt' Each vehicle 
with a GVWR. of 10.000 pound, or ; 
shall be capable of meeting the requii-- 
ments of anv applicable barrier crash te-t 
lollowed bv a static rollover, wit!', out al- 
teration of the vehicle during the tcV 
sequence. A particular vehicle need pc t 
meet further requirements after having 
been subjected to a single barrier crash 
test and a static rollover test. 

56.1 Frontal barrier crash When the 
vehicle traveling longitudinally forward 
at any speed up u> and including 30 mph 
impacts a fixed collision barrier that is 
perpendicular to the line of travel of the 
vehicle, or at any angle up to 30’ m 
either direction from the perpendicular 
to the line cf travel of the vehicle, with 
50th- percentile test dummies as specified 
in Part 572 of this chapter at each front 
outboard designated seating position and 
at any other position whose protection 
system is required to be tested b\ a dum- 
my under the provisions of Standard No 
208, under the applicable conditions of 
S7, fuel spillage shall not exceed the lim- 
its of S5.5, 

56.2 Rear moving barrier crash. When 
the vehicle is impacted from the rear by 
a barrier moving at 30 mph, with test 
dummies as specified in Part 572 of this 
chapter at each front outboard desig- 
nated seating position, under the appli- 



cable conditions cf S7, fuel spillage shall 
not exceed the limits of S5.5. 

S6.3 Lateral moving barrier crash. 
When the vehicle is impacted laterally on 
either side by a bairier moving at 20 mph 
with 50th-percentiie test dummies as 
specified in part 672 of this chapter at 
positions required for testing to Stand- 
ard No. 203, under the applicable condi- 
tions of 37. fuel -pillage shall not exceed 
the limits of S5.3. 

Sfi 4 Static reliever. When the ' ehicle 
is rotated on its longitudinal axis to each 
successive ir.cren n! of SO", following an 
impact crash Sfi.i, SC. 2, or S6.3. fuel spil- 
lage shall not preeed the limits of S5.6. 

56 6 Movin'? contoured barrier crash. 
When the moving contoured barrier as- 
sembly traveling longitudinally forward 
at any speed ur. to and including 30 mph 
impart? the test chicle (school bus with 
a GVWR exceeding 5 0.900 pounds' at any 
pain! and :ir;V. under the applicable 
conditions cf S7.I and S7.3. fuel spillage 
shall ne t ex: . c:i me Hmils of S3 5 

S7. Test conditions. The requirements 
of E5 and S5 ska'l be met under tne fol- 
lowing conditions, Where a range of con- 
ditions is sperifi d. the vehicle must be 
capabJ: of meeting the requirements at 
all prints within the range. 

57 i General . •■' e ndii ons. T he fol- 
low j eon dm- n apply to all tests. 

SI l.l The fuel tank is filled to any 
h.G from 99 to «:•, percent of c. parity 
wuh Stoddard o ... m. having the phys- 
ical a:u! ch' 3 : i.. - . ' properties of tvpe 1 
sol - ■:■ . 1 . Tab; 1 -TM Standard D484- 
71. "Stan.i.n c s. : catii ns for llydro- 
corb. n Dr. C.ea: :n Solvents ” 

S7 1.2 The fuel system other than the 
fuel tank h fill, cl with Stoddard solvent 
to its normal operating level. 

ST 1.3 In meeting the requirements of 
SG.l through S6.3. if the vehicle has an 
electrically driven fuel pump that nor- 
mally runs when the vehicle’s electrical 
system is activate I. it is operating at the 
time cf the barrier crash. 

S7 ) 1 The parking brake is disengaged 
and the trrnsmis~io;i is in neutral, ex- 
cept tnat in meeting the requirements 
of S6.5 the parking brake is set. 

37.1 5 Tires are inflated to manu- 
facturer’s specifications. 

S7.3.6 The vehicle, including test de- 
vices and instrumentation, is loaded as 
follows: 

; .a i Except as specified in S7.1.1. a pas- 
senger car is loaded to its unloaded ve- 
hicle weight plus its rated cargo and lug- 
gage capacity weight, secured in the lug- 
gage area, p.us the necessary test dum- 
mies as specified in S6 restrained only 
by means that are installed in the ve- 
hicle tor protection at its seating position. 

<bi Except as specified in S7 1.1, a 
multipurpose passenger vehicle, truck, or 
bus with a GVWR of 10,000 pounds or 
less is loaded to its unloaded vehicle 
weight, plus the necessary test dummies, 
as specified in S6., plus 300 pounds or its 
rated cargo and luggage capacity weight, 
whichever is less, secured to the vehicle 
and distributed so that the weight on 
each axle as measured at the tire-ground 
interface is in proportion to its GAWR. 



If the weight on any axle, when the ve- 
liicle is loaded to unloaded vehicle weight 
plus dummy weight, exceeds the axle’s 
proportional share of the test weight, the 
remaining weight shall be placed so that 
the weight on that axle remains the 
same. Each dummy shall be restrained 
only by means that are installed in the 
vehicle for protection at its seating posi- 
tion. 

ic) Except as specified in S7.1.1, a 
schoolbus with a GVWR greater than 
10,000 pounds is loaded to Its unloaded 
vehicle weight, plus 120 pounds of un- 
secured weight at each designated seat- 
ing position. 

57.2 Lateral moving barrier crash test 
conditions. The lateral moving barrier 
crash test conditions are those specified 
in S8.2 of Standard No. 208, 49 CFR 
571.203. 

57.3 Rear moving barrier test condi- 
tions. The rear moving barrier test con- 
ditions are those specified in S8.2 of 
Standard No. 208, 49 CFR 571.208, ex- 
cept for the positioning of the barrier 
and the vehicle. The barrier and test ve- 
hicle are positioned so that at impact — 

(a) The vehicle is at rest in its normal 
attitude: 

<b' The barrier is traveling at 30 mph 
with its face perpendicular to the longi- 
tudinal centerline of the vehicle: and 

(c> A vertical plane through the geo- 
metric center cf the barrier impact sur- 
face and perpendicular to that surface 
coincides with the longitudinal center- 
line of the vehicle. 

57.4 Static rollover test conditions. 
The vehicle is rotated about its longi- 
tudinal axis, with the axis kept horizon- 
tal, to each successive increment of 90°. 
ISO 1 , and 270 c at a uniform rate, with 90° 
of rotation taking place in any time in- 
terval from 1 to 3 minutes. After reach- 
ing each 90 s increment the vehicle is 
held in that position for 5 minutes. 

S7 5 Moving contoured barrier test 
conditions. The following conditions ap- 
ply to the moving contoured barrier 
crash test. 

57.5.1 Tire moving barrier, which is 
mounted on a carriage as specified in 
figure 1, is of rigid construction, sym- 
metrical about a vertical longitudinal 
plane. The contoured impact surface, 
which is 24.75 inches high and 78 inches 
wide, conforms to the dimensions shown 
in figure 2, and is attached to the car- 
riage as shown in that figure. The ground 
clearance to the lower edge of the impact 
surface is 5.25 it 0.5 Inches. The wheel- 
base is 120 -t 2 inches. 

57.5.2 The moving contoured barrier, 
including the Impact surface, supporting 
structure, and carriage, weighs 4,000 
it 50 pounds with the weight distributed 
so that 900 ± 25 pounds is at each rear 
wheel and 1100 ± 25 pounds is at each 
front wheel. The center of gravity is lo- 
cated 54.0 it 1.5 inches rearward of the 
front wheel axis, In the vertical longi- 
tudinal plane of symmetry, 15.8 inches 
above the ground. The moment of in- 
ertia about the center of gravity Is: 

1 M = 27 ± 13.6 slug ft 3 
/. =3475 ± 174 slug ft 3 



FEDERAL REGISTER, VOL. 40, NO. 200 — WEDNESDAY, OCTOBER 15, 1975 



C-4 



RULES ANu> REGULATIONS 

57.5.5 Th® concrete surface upon 
which the vehicle Is tested is level, rigid, 
and of uniform construction, with a skid 
number of 75 when measured in accord- 
ance with American Society of Testing 
and Materials Method E-274-85T at 40 
mph, omitting water delivery as speci- 
fied in paragraph 7.1 of that method. 

57.5.6 The barrier assembly is re- 
leased from the guidance mechanism im- 
mediately prior to impact with the 
vehicle. 




57.5.3 The moving contoured barrier 
has a solid nonsteerable front axle and 
fixed rear axle attached directly to the 
frame rails with no spring or other type 
of suspension system on any wheel. (The 
moving barrier assembly is equipped with 
a braking device capable of stopping its 
motion.) 

57.5.4 The moving barrier assembly is 
equipped with G78-15 pneumatic tires 
with a tread width of 6.0 ± 1 inch, in- 
flated to 24 psi. 



FIG I -COMMON CARRIAGE FOR MOVING BARRIERS 






uoMom or *4onu 

f, - 371 ± 0.6 StUG • FT' 06? * IM ko ■ •’> 

Iff - DITJ ± 174 V UO FT* 1471 1 * 234 kg • 

OTIS 

I uftft FUAMt OMDU X 0.2) tn WALL (M2 mm 01 A X * mm WMAJ IFEh TUttMO 




Jtotffc- SADtSi. 

i. lO»TIHANft.6M5U XfiOtN WAUCI3J O mm WAUJ TtMfHC*. 

3. FAQ naff 073 M |lt mmt TMICV COL* tOuTO JTOi 

4, ifA0*4G toot 1.0 X 4.0 I* 17> X IC2 — i STTBL 9AM0. IHA* C*G4i WOKiH 

i. Mi INHEt OHF04C4M4MI5 4 JO X 1.0 X O.lf 1*001 X »l K5«l TW4ING. 



nc. COMMON CARRIAGE WITH CONTOURED IMPACT 
SURFACE ATTACHEO 



[PR Doc.76-27612 Piled i0-14-75;8:45 anal 



48355 







FEOBtAt 8&SiSTES, VOL. 40, MO. 300— WIIM4ESOAV, OCTOBB* 15, 1075 

C-5 



APPENDIX D 

FEDERAL MOTOR VEHICLE 
SAFETY STANDARD NO. 208 



D-l 



Effective: January 1, 1972 
September 1, 1973 
August IS, 1975 
August 15, 1977 



MOTOR VEHICLE SAFETY STANDARD NO. 208 

Occupant Crash Protection in Passenger Cars, Multipurpose Passenger 

Vehicles, Trucks and Buses 

(Docket No. 69-7; Notice No. 9) 



51. Scope. This standard specifies perform- 
ance requirements for the protection of vehicle 
occupants in crashes. 

52. Purpose. The purpose of this standard is 
to reduce the number of deaths of vehicle oc- 
cupants, and the severity of injuries, by specify- 
ing vehicle crashworthiness requirements in terms 
of forces and accelerations measured on anthro- 
pomorphic dummies in test crashes, and by 
specifying equipment requirements for active 
and passive restraint systems. 

53. Application. [This standard applies to 
passenger cars, multipurpose passenger vehicles, 
trucks, and buses. In addition, S9, Pressure 
vessels and explosive devices , applies to vessels 
designed to contain a pressurized fluid or gas, 
and to explosive devices, for use in the above 
types of motor vehicles as part of a system de- 
signed to provide protection to occupants in the 
event of a crash. (37 F.R. 9222 — May 6, 1972. 
Effective: 6/2/72)] 

$4. General requirements. 

54.1 Passenger cars. 

[S4.1.1 Passenger cars manufactured from Jan- 
uary 1, 1972, to August 31, 1973. Each passenger 
car manufactured from January 1, 1972, to 
August 31, 1973, inclusive, shall meet the require- 
ments of S4.1.1.1, S4.1.1.2, or S4.1.1.3. A pro- 
tection system that meets the requirements of 

S4.1.1.1 or S4.1.1.2 may be installed at one or 
more designated seating positions of a vehicle 
that otherwise meets the requirements of 
S4.1.1.3. (38 F.R. 21930— August 14, 1973. Ef- 

fective: 8/31/73)] 

54. 1.1.1 First option — complete passive protec- 
tion system'. The vehicle shall meet the crash 

(Rev. 8/10/73) 



protection requirements of S5 by means that re- 
quire no action by vehicle occupants. 

S4.1.1.2 Second option — lap belt protection 
system with belt warning. The vehicle shall — 

(a) [At each designated seating position have 
a Type 1 seat belt assembly or a Type 2 seat belt 
assembly with a detachable upper torso portion 
that conforms to S7.1 and S7.2 of this standard. 
(37 F.R. 3911 — February 24, 1972. Effective: 
2/24/72)] 

(b) At each front outboard designated seating 
position have a seat belt warning system that 
conforms to S7.3 ; and 

(c) Meet the frontal crash protection require- 
ments of S5.1, in a perpendicular impact, with 
respect to anthropomorphic test devices in each 
front outboard designated seating position re- 
strained only by Type 1 seat belt assemblies. 

S4.1.1.3 Third option — lap and shoulder belt 
protection system with belt warning. 

S4. 1.1 .3.1 Except for convertibles and open- 
body vehicles, the vehicle shall — 

(a) At each front outboard designated seat- 
ing position have a Type 2 seatbelt assembly 
that conforms to Standard No. 209 and S7.1 and 
S7.2 of this standard, with either an integral or 
detachable upper torso portion, and a seatbelt 
warning system that conforms to S7.3; 

(b) At each designated seating position other 
than the front outboard positions, have a Type 1 
or Type 2 seat belt assembly that conforms to 
Standard No. 209 and to S7.1 and S7.2 of this 
standard; and 

(c) When it perpendicularly impacts a fixed 
collision barrier, while moving longitudinally 



PART 571; S 208-1 



D-2 



Effective: 1/1/72; 9/1/73; 

6/15/75; 8/15/77 



forward at any speed up to and including 30 
n.p.h., under the test conditions of S8.1 with 
anthropomorphic test devices at each front out- 
board position restrained by Type 2 seatbelt as- 
semblies, experience no complete separation of 
any load-bearing element of a seatbelt assembly 
or anchorage. 

54.1.1.3.2 Convertibles and open-body type 
vehicles shall at each designated seating position 
have a Type 1 or Type 2 seatbelt assembly that 
conforms to Standard No. 209 and to S7.1 and 

S7.2 of this standard, and at each front outboard 
designated seating position have a seatbelt warn- 
ing system that conforms to S7.3. 

54.1 .2 [Passenger cars manufactured from 
September 1, 1973, to August 31, 1976, Pas- 
senger cars manufactured from September 1, 
1973, to August 31, 1976, inclusive, shall meet 
the requirements of S4.1.2.1, S4.1.2.2, or S4. 1.2.3. 
A protection system that meets the requirements 
of S4.1.2.1 or S4.1.2.2 may be installed at one or 
more designated seating positions of a vehicle 
that otherwise meets the requirements of S4.1.2.3. 
(40 F.R. 33977— August 13, 1975. Effective: 
8/13/75] * 

54. 1.2.1 First option— complete passive protec- 
tion system. The vehicle shall meet the crash 
protection requirements of S5 by means that re- 
quire no action by vehicle occupants. 

54. 1.2.2 Second option — head-on passive pro- 
tection system. The vehicle shall — 

£(a) At each designated seating position have 
a Type 1 seat belt assembly or a Type 2 seat belt 
assembly with a detachable upper torso portion 
that conforms to S7.1 and S7.2 of this standard. 
(37 F.R. 3911— February 24, 1972. Effective: 
2/24/72)] 

(b) At each front designated seating position, 
meet the frontal crash protection requirements 
of S5.i, in a perpendicular impact, by means that 
require no action by vehicle occupants; 

(c) At each front designated seating position, 
meet the frontal crash protection requirements of 
S5.1, in a perpendicular impact, with a test de- 
vice restrained by a Type 1 seatbelt assembly ; 
and 



(d) At each front outboard designated seating 
position, have a seatbelt warning system that 
conforms to S7.3. 

S4.1.2.3 Third option- — lap and shoulder belt 
protection system with belt warning. 

S4.1 .2.3.1 [Except for convertibles and open- 
body vehicles, the vehicle shall — 

(a) At each front outboard designated seat- 
ing position have a seat belt assembly that con- 
forms to S7.1 and S7.2 of this standard, and a 
seat belt warning system that conforms to S7.3. 
The belt assembly shall be either a Type 2 seat 
belt assembly with a nondetachablo shoulder belt 
that conforms to Standard No. 209 (§ 571.209), 
or a Type 1 seat belt assembly such that with a 
test device restrained by the assembly the ve- 
hicle meets the frontal crash protection require- 
ments of S5.1 in a perpendicular impact. 

(b) At any center front designated seating 
position, have a Type 1 or Type 2 seat belt as- 
sembly that conforms to Standard No. 209 
(§ 571.209) and to S7.1 and S7.2 of this standard, 
and a seat belt warning system that conforms to 
S7.3; and 

(c) At each other designated seating position, 
have a Type 1 or Type 2 seat belt assembly that 
conforms to Standard No. 209 (§ 571.209) and 

S7.1 and S7.2 of this standard. (39 F.R. 38380 — 
October 31, 1974. Effective: 10/29/74)] 

54.1 .2.3.2 [Convertibles and open-body type 
vehicles shall at each designated seating position 
have a Type 1 or Type 2 seat belt assembly that 
conforms to Standard No. 209 (§ 571.209) and 
to S7.1 and S7.2 of this standard, and at each 
front designated seating position have a seat belt 
warning system that conforms to S7.3. (39 F.R. 
38380— October 31, 1974. Effective: 10/29/74)] 

S4.1.3 [Reserved. (40 F.R. 33977— August 13, 
1975. Effective: 8/13/75)] 

54.2 Trucks and multipurpose passenger ve- 
hicles with GVWR of 10,000 pounds ©r less. 

S4.2.1 [Trucks and multipurpose passenger 
vehicles, with GVWR of 10,000 pounds^ or less, 
manufactured from January 1, 1972, to December 
31, 1975. Each truck and multipurpose pas- 
senger vehicle with a gross vehicle weight rating 



tRov. 9/3/rsi PART 571; S 208-2 

Changed by 41F.R. 36494 D-3 



Effective: 1/1/72; 8/15/73; 

8/15/75; 8/15/77 



of 10,000 pounds or less, manufactured from 
January 1, 1972, to December 31, 1975, inclusive, 
shall meet the requirements of S4.2.1.1 or S4.2.1.2, 
or at the option of the manufacturer, the re- 
quirements of S4.2.2. A protection system that 
meets the requirement of S4.2.1.1 may be in- 
stalled at one or more designated seating posi- 
tions of a vehicle that otherwise meets the 
requirements of S4.2.1.2. (40 F.R. 28805 — July 

9,1975. Effective: 7/9/75)] 

54.2.1.1 First option — complete passive pro- 
tection system. The vehicle shall meet the crash 
protection requirements of S5 by means that re- 
quire no action by vehicle occupants. 

54.2.1.2 Second option — belt system. The ve- 
hicle shall have seat belt assemblies that conform 
to Standard 209 installed as follows: 

(a) A Type 1 or Type 2 seat belt assembly 
shall be installed for each designated seating posi- 
tion in convertibles, open-body type vehicles, and 
walk-in van-type trucks. 

(b) In all vehicles except those for which re- 
quirements are specified in S4.2.1.2(a), a Type 
2 seat belt assembly shall be installed for each 
outboard designated seating position that in- 
cludes the windshield header within the head 
impact area, and a Type 1 or Type 2 seat belt 
assembly shall be installed for each other desig- 
nated seating position. 

54.2.2 |[Trucks and multipurpose passenger ve- 

hicles, with GVWR of 10,000 pounds or less, 
manufactured from January 1, 1976, to August 14, 
1977. Each truck and multipurpose passenger 
vehicle, with a gross vehicle weight rating of 
10,000 pounds or less, manufactured from Jan- 
uary 1, 1976, to August 14, 1977, inclusive, shall 
meet the requirements of S4.1.2 (as specified for 
passenger cars), except that forward control ve- 
hicles, convertibles, open-body type vehicles, 
walk-in van-type trucks, motor homes, and ve- 
hicles carrying chassis-mount campers may in- 
stead meet the requirements of S4.2.1.2. (40 

F.R. 28805 — July 9, 1975. Effective: 7/9/75)] 

54.2.3 Trucks and multipurpose passenger ve- 
hicles, with GVWR of 10,000 pounds or less, 
manufactured on or after August 15, 1977. Each 
truck and multipurpose passenger vehicle, with 
a gross vehicle weight rating of 10,000 pounds 



or less, manufactured on or after August 15, 1977, 
shall meet the occupant crash protection require- 
ments of S5 by means that require no action by 
vehicle occupants, except that forward control 
vehicles may instead meet the requirements of 
S4.2.1.2, and convertibles, open-body vehicles, 
walk-in van-type trucks, motor homes, and ve- 
hicles carrying chassis-mounted campers may in- 
stead meet the requirements of S4.1.2.2. 

54.3 Trucks and multipurpose passenger ve- 
hicles, with GVWR of more than 10,000 pounds. 
Each truck and multipurpose passenger vehicle, 
with a gross vehicle weight rating of more than 
10,000 pounds, manufactured on or after Janu- 
ary 1, 1972, shall meet the requirements of 
S4.3.1 or S4.3.2. A' protection system that meets 
the requirements of S4.3.1 may be installed at one 
or more designated seating positions of a vehicle 
that otherwise meets the requirements of St. 3.2. 

54.3.1 First option — complete passive protec- 
tion system. The vehicle shall meet the crash 
protection requirements of S5 by means that re- 
quire no action by vehicle occupants. 

54.3.2 Second option — belt system. The ve- 
hicle shall, at each designated seating position, 
have either a Type 1 or a Type 2 seatbelt as- 
sembly that conforms to Standard No. 209. 

54.4 Buses. Each bus manufactured on or 
after January 1, 1972, shall meet the require- 
ments of S4.4.1 or S4.4.2. 

54.4.1 First option — complete passive protec- 
tion system — driver only. The vehicle shall meet 
the crash protection requirements of S5, with re- 
spect to an anthropomorphic, test device in the 
driver’s designated seating position, by means 
that require no action by vehicle occupants. 

54.4.2 Second option — belt system — driver only. 
The vehicle shall, at the driver’s designated seat- 
ing position, have either a Type 1 or a Type 2 
seatbelt assembly that conforms to Standard No. 
209. 

54.5 Other general requirements. 

S4.5.1 Labeling and driver’s manual informa- 
tion. [Each vehicle shall have a label settin 
forth the manufacturer's recommended sohedul 
for the maintenance or replacement, necessary 
to retain the performance required by this stand- 
ard, of any crash deployed occupant protection 
system. The schedule shall be specified by month 



(Rev. 9/3/75) 



PART 571; S 208-3 
D-4 



tt ^ 



Effective: 1/1/72; 8/15/73; 

6/15/75; 8/15/77 



and year, or in terms of vehicle mileage, or by 
•'terval s measured from the date appearing on 
«ue vehicle < ortification label provided pursuant 
to 49 CFR rart 5 07. The label shall be perma- 
nently affixed to the vehicle within the passenger 
compartment and lettered in English in block 
capitals and numerals not less than three thirty- 
seconds of an inch high. Instructions concern- 
ing maintenance or replacement of a system and 
a description of the functional operation of the 
system shall be provided with each vehicle, with 
an appropriate reference on the label. If a 
vehicle owner’s manual is provided, this infor- 
mation shall be included in the manual. 
(39 F.R. 1513 — January 10, 1974. Effective: 
1/10/74)3 

54.5.2 Readiness indicator. £An occupant 
protection system that deploys in the event of a 
crash shall have a monitoring system with a 
readiness indicator. The indicator shall monitor 
its own readiness and shall be clearly visible 
from the driver’s designated seating position. 
A list of the elements of the system being moni- 
tored by the indicator shall be included with the 
information, furnished in accordance with S4.5.1 
but need not be included on the label. (36 F.R. 
>254 — October 1, 1971. Effective: 1/1/72)3 

£54.5.3 Passive belts. Except as provided in 
S4.5.3.1, a seat belt assembly that requires no 
action by vehicle occupants (hereinafter referred 
to as a “passive belt”) may be used to meet the 
crash protection requirements of any option un- 
der S4 and in place of any seat belt assembly 
otherwise required by that option. 

54.5.3.1 A passive belt that provides only 
pelvic restraint may not be used pursuant to 
S4.5.3 to meet the requirements of an option that 
requires a Type 2 seat belt assembly. 

54.5.3.2 A passive belt, furnished pursuant to 
S4.5.3, that provides both pelvic and upper torso 
restraint may have either a detachable or non- 
detachablo upper torso portion, notwithstanding 
provisions of the option under which it is fur- 
nished. 

54.5.3.3 [A passive, belt furnished pursuant 
to S4.5.3 shall — 

(a) Conform to S7.1 and S7.2 of this stand- 
ard; and 

«v. 12/2/74) 



(b) In place of a warning system that con- 
forms to S7.3 or 87.5a of this standard, be 
equipped with a warning system as specified in 
subparagraph (1), except that a scat licit as- 
sembly provided in a vehicle that is manufac- 
tured prior to February 24, 1975, may, at the 
option of the manufacturer, be equipped with a 
warning system as specified in subparagraph (1) 
or as specified in subparagraph (2) : 

(1) At the left front designated seating posi- 
tion (driver’s position), be equipped with a warn- 
ing system that activates, for a period of not 
less than 4 seconds and not more than 8 seconds 
(beginning when the vehicle ignition switch is 
moved to the “on” or the “start.” position), a 
continuous or flashing warning light, visible to 
the driver, displaying the words “Fasten Seat 
Belts” or “Fasten Belts” when condition (A) 
exists, and a continuous or intermittent, audible 
signal when condition (A) exists simultaneously 
with condition (B). 

(A) The vehicle’s ignition switch is moved 
to the “on” position or to the “start” position. 

(B) The driver’s lap belt is not in use, as de- 
termined by the belt latch mechanism not being 
fastened. 

(2) Be equipped with a warning system that 
activates, for at least, one minute, a continuous 
or intermittent audible signal and a continuous 
or flashing warning light, visible to the driver, 
displaying the words “Fasten Seat Belts” or 
“Fasten Belts”, whenever the ignition switch is 
in the “start” position and the latch mechanism 
is not fastened, and whenever the vehicle engine 
is running, the transmission gear selector is 
placed in any forward position, and the latch 
mechanism is not fastened. (39 F.R. 42692 — 
December 6, 1974. Effective: 12/3/74)3 

54.5,3.4 A passive belt furnished pursuant to 
S4.5.3 that is not required to meet, the perpen- 
dicular frontal crash protection requirements of 
S5.1 shall conform to the webbing, attachment 
hardware, and assembly performance require- 
ments of Standard No. 209. (36 F.R. 23725— 

December 14, 1971. Effective: 1/1/72)3 

S5. Occupant crash protection requirements. 

S5.1 Frontal barrier crash. JBVhen the vehicle, 
traveling longitudinally forward at any speed 
up to and including 30 m.p.h., impacts a fixed 



PART 571; S 208-4 



D-5 



Effective: 1/1/72; 8/15/73; 

8/15/75; 8/15/77 



collision barrier that is perpendicular to the line 
of travel of the vehicle, or at any angle up to 
30° in either direction from the perpendicular 
to t tie line of travel of the vehicle, under the 
applicable conditions of S3, with anthropo- 
morphic test devices a t each designated seating 
position for which a barrier crash test is re- 
quired under S4, it shall meet the injury criteria 
of SG. (37 F.R. 3911 — February 24, 1972. Ef- 
fective: 2/24/72)3 

$5,2 Lateral moving barrier crash. When the 

vel ]■:• impacted laterally on either side by 
m r am ;no% :ng at 20 m.p.h., with test devices 
at the outboard designated seating positions ad- 
jacent to the impacted side, under the applicable 
conditions of S8, it shall meet the injury criteria 
of SG. 

S5.3 Rollover. 'When the vehicle is subjected 
to a rollover test in either lateral direction at 
30 m.p.h. with test devices in the outboard desig- 
nated seating positions on its lower side as 
mounted on the tit platform, under the appli- 
cable conditions of S3, it shall meet the injury 
criteria of S6.1. [However, vehicles manufac- 
tured before August 15, 1977, that conform to 
the requirements of Standard No. 216 (jj 571.216) 
need not conform to this rollover test require- 
ment (36 F.R. 23299 — December 3, 1971. Effec- 
tive: 1/1/72)3 

S.6 Injury criteria. 

56.1 All portions of the test device shall be 
contained within the outer surfaces of the ve- 
hicle passenger compartment throughout the test. 

56.2 [The resultant acceleration at the center 
of gravity of the head shall be such that the 
expression : 




shall not exceed 1,000, where a is the resultant 
acceleration expressed as a multiple of <j (the 
acceleration of gravity), and t, and t 2 are any 
two points in time during the crash. However, 
in the case of a passenger car manufactured be- 
fore August 31, 1976, or a truck or multipurpose 
passenger vehicle with a GVAVR of 10,000 pounds 
or less manufactured before August 15, 1977, 

(Rov. 8/8/751 



when the dummy is restrained by a seat belt 
system, t, and t 2 are any two points in time dur- 
ing any interval in which the head is in con- 
tinuous contact with a part of the vehicle other 
than the belt system. (40 F.R. 33977 -August 
13, 1975. infective : 8/13/75)3 

$6.3 [T1 le resultant acceleration at the center 
of gravity of the upper thorax shall not exceed 
60g, except for intervals whose cumulative, dura- 
tion is not more than 3 milliseconds. However, 
in the case of a passenger car manufactured 
before Aueiist 31, 1976, or a truck or multipur- 
pose passenger vehicle with a GVWR of 10,000 
pounds or less manufactured before August 15, 
1977, the resultant acceleration at the center of 
gravity of the upper thorax shall be such that 
the severity index calculated by the method de- 
scribed in SAE Information Report J885a, 
October, 1906, shall not exceed 1,000. (40 F.R. 

33977— August 13, 1975. Effective: 8/13/75)3 

S6.4 [The force transmitted axially through 
each upper eg shall not exceed 1,700 pounds. 
(37 F.R. 21903 — November 23, 1972. Effective: 
11/23/72)3 

S7. Seal belt assembly requirements — passenger 
cars. 

57.1 Adjustment. 

57.1.1 Except as specified in S7.1.1.1 and 

S7.1.1.2, the lap belt of any seat belt assembly 
furnished in accordance with S4.1.1 and S4.1.2 
shall adjust by means of an emergency-locking 
or automatic locking retractor that conforms to 
Standard No. 209 to lit persons whose dimen- 
sions range from those of a 50th-perccntile 6- 
year-old child to those of a 95th-percentile adult 
male and the upper torso restraint shall adjust 
by means of an emergency-locking retractor or 
a manual adjusting device that conforms to 
Standard No. 209 to (it persons whose dimensions 
range from those of a 5th percentile adult female 
to tiiose of a 95th-percentile adult male, with 
t he scat in any position and t he seat back in the 
manufacturer’s nominal design riding position. 
[However, an upper torso restraint furnished in 
accordance with S4.1.2.3.1 (a) shall adjust by 
means of an emergency-locking retractor that 
conforms to Standard No. 209. (37 F.R. 3911 — 

February 24, 1972. Effective: 2/24/72)3 



PART 571 ; S 208-5 



D-6 



Effective: 1/1/72; 8/15//3; 

8/15/75; 8/15/77 



57.1.1.1 A scat holt assembly installed at the 
driver’s seating position shall adjust to fit per- 
sons whose dimensions range from those of a 
5th-perrentile adult female to those of a Oath- 
percentile adult male. 

57.1.1.2 A seat belt assembly installed at any 
designated seating position other than the out- 
board positions of the front and second seats 
shall adjust either by a retractor as specified in 
S7.1.1 or by a manual adjusting device that con- 
forms to Standard No. 209. 

57.1.2 The intersection of the upper torso belt 
with the lap belt in any Type 2 seat belt assembly 



furnished in accordance with Si. 1.1 or Si. 1.2, 
with the upper torso manual adjusting device, 
if provided, adjusted in accordance with the 
manufacturer’s instructions, shall be at least G 
inches from the front vertical centerline of a 
oOth-percenf ile adult male occupant, measured 
along the centerline of the lap belt, with the seat 
in its rearmost and lowest adjustable position and 
with the scat back in the manufacturer’s nomi- 
nal design riding position. 

S7.1.3 The weights and dimensions of the ve- 
hicle occupants specified in this standard are as 
follows : 





50th-percentile 
6-year-old child 


5th-percentilc 
adult female 


50th-percentilc 
adult male 


95th-percontile 
adult male 


Weight 


47.3 pounds 


. 102 pounds 


1G4 pounds 


.215 poun ds. 


Erect sitting height. 


25. 4 inches 


. 30. 9 inches 


35. 7 inches 


. 38 inches. 


Hip breadth (sitting) 


8. 4 inches 


. 12. 8 inches 


14.5 inches 


. 16. 5 inches. 


Hip circumference (sitting) 


23.9 inches 


. 30. 4 inches 


42 inches 


. 47. 2 inches. 


Waist circumference (sitting) 


20. 8 inches 


. 23. 6 inches 


33 inches 


. 42. 5 inches. 


Chest depth.. 

Chest circumference: 

(nipple) 




7. 5 inches 

. 30. 5 inches 


. 9 inches 

•---] 


. 10. 5 inches. 


(upper)... 

Gower) 




. 29. 8 inches 

_ 26. 6 inches 


37. 7 inches. 


. . 44. 5 inches. 



S7.2 Latch mechanism. A seat belt assembly 
installed in a passenger car shall have a latch 
mechanism — 

(a) Whose components are accessible to a 
seated occupant in both the stowed and opera- 
tional positions; 

[(b) That releases both the upper torso re- 
straint and the lap belt simultaneously, if the 
assembly has a lap belt and an upper torso re- 
straint that require unlatching for release of the 
occupant ; and (39 F.R. 14593 — April 25, 1974. 
Effective: 5/27/74)3 

(c) That releases at a single point by a push- 
button action. 

S7.3 Seat belt warning system. [A seat belt 
assembly provided in accordance with S4.1 shall 
be equipped with a seat belt warning as specified 
in S7.3a, except that a seat belt assembly pro- 
vided in accordance with S4.1 in a vehicle manu- 
factured prior to February 24, 1975, may, at the 
option of the manufacturer, be equipped with 
either a seat belt warning as specified in S7.3.1 
through S7.3.5 or a seat belt, warning as specified 
in S7.3a. (39 F.R. 42G92 — December G, 1974. 

Effective: 12/3/74)3 



S7.3.1 [[Seat belt assemblies provided at the 
front outboard seating positions in accordance 
with S4.1.1 or S4.1.2 shall have a warning system 
that activates, for at least one minute, a con- 
tinuous or intermittent audible signal and con- 
tinuous or flashing warning light, visible to the 
driver, displaying the words ‘‘Fasten Seat Belts” 
or “Fasten Belts” when condition (a) exists 
simultaneously with either of conditions (b) or 
(c). 

[(a) The vehicle's engine is operating and the 
transmission gear selector is in any forward po- 
sition. (36 F.R. 23725 — December 14, 1971. 
Effective: 1/1/72)3 

(b) [The driver’s lap belt is not in use, as 
determined, at the manufacturer’s option, either 
by the belt latch mechanism being fastened or 
by the belt being extended at least 4 inches from 
its stowed position. (37 F.R. 3911 — February 
24, 1972. Effective: 2/24/72)3 

(c) [A person of at least the weight of a 50th 
percentile adult male is seated with the belt fast- 
ened at the driver’s position, and a person of at 
least the weight of a 50th percentile 6-year-old 
child is seated in the right front designated seat- 



(Rev. 12/2/74) PART 571; S 208-6 

D-7 



Effective; 1/1/72; 8/15/73; 

8/15/75; 8/15/77 



ing position and the lap boll for that position 
is not in use, as determined, at the manufacturers 
option, either by the belt latch mechanism being 
fastened or by the belt being extended at least 
4 inches from its stowed position. (37 F.K. 
3011 — February 24, 1972. Effective: 2/24/7.2)3 

57.3.2 The warning system shall either — 

[(a) Not activate when the lap belt at each 

occupied front outboard seating position is ex- 
tended to any length greater than the length 
necessary to fit a 50th-percentile 6-year-old child 
when the seat is in the rearmost and lowest ad- 
justment position; 

(b) Not activate when the lap belt at each 
occupied front outboard position is buckled: or 

(c) Not activate when the operation specified 
in (a) or (b) is performed at each occupied 
front outboard seating position after the occu 
pant is seated. (37 F.K. 132005 — July G, 1072. 
Effective: 1/1/73)3 

57.3.3 [The warning systems shall not acti- 
vate if the vehicle has an automatic transmission, 
tho engine is operating, and the gear selector is 
in the “Park” position. (37 F.K. 391] -Feb- 
ruary 24, 1972. Effective: 2/24/72)3 

57.3.4 [Notwithstanding the provisions of 
S7.3.1 and S7.3.5.2, when tho engine of a vehicle 
with a manual transmission is operating, the 
warning system shall either — 

(a) Not activate when the transmission is in 
neutral; or 

(b) Not activate when the parking brake is 
engaged. 

57.3.5 [The above provisions of S7.3 shall 
apply to seat belt assemblies furnished in accord- 
ance with S4. 1.2.3, with the following exceptions: 
(39 F.R. 38380— October 31, 1974. Effective: 
10/29/74)3 

$7. 3. 5.1 The warning system shall also he pro- 
vided for the center front seating position, if any. 

57.3.5.2 In addition to the conditions specified 
in S7.3.1, the warning system shall activate if — 

(a) The vehicle’s engine is operating and the 
transmission gear selector is in any forward po- 
sition, and 

(b) A person of at least the weight of a 50th 
percentile adult male is seated with the belt fast- 
ened at the driver’s position, and a person of at 
least the weight of a 5th percentile adult female 

(R«v. 12/2/74) 



is seated in a center front designated seating 
position and the lap belt for the center front 
position is not in use, as determined, at the manu- 
facturers option, cither by the belt latch mech- 
anism bo ng fastened or the belt being extended 
at least 1 inches from its stowed position. 

57.3.5.3 'Flic provisions of S7.3.2 shall apply 
to all front seating positions. 

57.3.5.4 [Notwithstanding the other provi- 
sions of S7.3, the warning system shall activate 
whenever the ignition switch is in the “start" 
position and the operation of (lie belt system 
at each occupied front outboard designated seat- 
ing position lias not been performed after the 
occupant is seated and condition (a) or (b) 
exists, licit operation for the. purpose of this 
requirement shall he, at the manufacturer’s op- 
tion, either the extension of the belt assembly 
at least -I inches from ils stowed position, or the 
fastening of the belt latch mechanism. 

(a) A person of at least, the weight of a 5th- 
percentilc adult, female is seated at the driver’s 
seating position. 

(b) A person of at least the. weight of a 50th- 
perccntile r hilt male is seated at the driver’s 
seating position and a person of at least the 
weight of a 50th-percentile G-year-old child is 
seated at the right front seating position. (39 
F.K. 38380 — October 31, 1974. Effective: 10/ 
29/74)] 

[S7.3o A scat belt, assembly provided at the 
driver's seating position shall be equipped with 
a warning system that activates, for a period of 
not less than -1 seconds and not more than 8 
seconds (beginning when the vehicle ignition 
switch is moved to the “on" or the “start” posi- 
tion), a continuous or flashing warning light, 
visible to the. driver, displaying the words 
“Fasten Feat Kelts" or “Fasten Kelts” when 
condition (a) exists, and a continuous or inter- 
mittent audible signal when condition (a) exists 
simultaneously with condition (b). 

(a) The vehicle’s ignition switch is moved to 
the “on" position or to the “start” position. 

(b) The driver's lap belt is not in use, as de- 
termined, at the oj)! ion of the manufacturers, 
either by the belt, latch mechanism not being 
fastened, or by the holt, not being extended at 
least 4 inches from its stowed position. (39 F.K. 
42092 — December 0, 1974. Effective: 12/3/74)3 



PART 571; S 208-7 

D-8 



Effective: 1/1/72; 6/15/73; 

8/15/75; 8/15/77 



S7.4 Belt interlock system. [Revoked. (39 
F.R. 38380 — October 31, 1974. Elective: 10/ 
29/74)3 

S8. Test conditions. 

58.1 General conditions. The following con- 
ditions apply to the frontal, lateral, and rollover 
tests. 

58.1.1 The vehicle, including test devices and 
instrumentation, is loaded as follows: 

(a) Passenger cars. A passenger car is loaded 
to its unloaded vehicle weight plus its rated 
cargo and luggage capacity weight, secured in 
the luggage area, plus the weight of the neces- 
sary anthropomorphic test devices. 

(b) Multipurpose passenger vehicles , trucks , 
and buses. [A multipurpose passenger vehicle, 
truck, or bus is loaded to its unloaded vehicle 
weight plus 300 pounds or its rated cargo and 
luggage capacity weight, whichever is less, se- 
cured in the load carrying area and distributed 
as nearly as possible in proportion to its gross 
axle weight ratings, plus the weight of the neces- 
sary anthropomorphic test devices. (36 F.R. 
19254 — October 1, 1971. Effective: 1/1/72)3 

58.1.2 Adjustable seats are in the adjustment 
position midway between the forwardmost and 
rearmost positions, and if separately adjustable 
in a vertical direction, are at the lowest position. 

58.1.3 Adjustable seat backs are in the manu- 
facturer’s nominal design riding position. 

58.1.4 Adjustable steering controls are ad- 
justed so that the steering wheel hub is at the 
geometric center of the locus it describes when it 
is moved through its full range of driving posi- 
tions. 

58.1.5 Movable vehicle windows and vents are 
in tho fully closed position. 

58.1.6 Convertibles and open-body type ve- 
hicles have the top, if any, in place in the closed 
passenger compartment configuration. 

58.1.7 Doors are fully closed and latched but 
not locked. 

58.1.8 [Anthropomorphic test devices used for 
the evaluation of restraint systems manufactured 
pursuant to sections S4. 1.2.1 and S4.1.2.2 con- 

IRev. 10/29/74) 



form to the requirements of Part 572 of this 
title. (38 F.R. 20419— August 1, 1973. Effec- 
tive: 8/15/73)3 

58.1.9 Each test device is clothed in form- 
fitting cotton stretch garments. 

58.1.10 [Limb joints are set at /</, barely 
restraining the weight of the limb when extended 
horizontally. Leg joints are adjusted with the 
torso in the supine position. (38 F.R. 20449 — 
August 1, 1973. Effective: 8/15/73)3 

58.1.11 Each test device is firmly placed in a 
designated seating position in the following 
manner: 

(a) The head is aligned by placing the test 
device on its back on a rigid, level surface and 
by adjusting the head so that it touches the level 
surface and is laterally centered with respect to 
the device’s axis of symmetry. 

(b) The test device is placed in the vehicle in 
the normal upright sitting position and a rigid 
roller, 6 inches in diameter and 24 inches long, 
is placed transversely as 1ow t as possible against 
the front of the torso. 

(c) The roller is pressed horizontally against 
the torso with a force of 50 pounds. 

(d) Force is applied at the shoulder level to 
bend the torso forward over the roller, flexing 
the lower back, and to return the test device to 
the upright sitting posture. 

(e) The roller is slowly released. 

58.1.12 Except as otherwise herein specified, 
the test devices are not restrained during impacts 
by any means that require occupant action. 

58.1.13 [The hands of the test device in the 
driver’s designated seating position are on the 
steering wheel rim at the horizontal centerline. 
The right foot rests on the undepressed accele- 
rator pedal, with the heel in contact with the 
point where the centerline of the upper surface 
of the undepressed accelerator pedal intersects 
the upper surface of the floor covering. The 
left leg is placed as in S8.1.14. (36 F.R. 19254 — 
October 1, 1971. Effective : 1/1/72)3 

58.1.14 The hands of each other test device are 
resting on the seat with the palms touching the 
legs, and the upper arms are resting against the 
seat back and flush with the body. Where pos- 
sible, tho legs are outstretched, with the thighs 



PART 571; S 208-8 



D-9 



Efledive: 1/1/72; 8/15/73; 

8/15/75; 8/15/77 



on the seat and the heels touching the floor with 
the foot at 90° to the tibia. Otherwise, the tibia 
are vertical with the feet resting on the floor. 
The left leg of a test device in the center front 
designated seating position is on the vehicle 
centerline, and the right leg is in the right foot- 
well. The left and right legs of a test device in 
the center rear designated seating position are 
in the left and right footwells, respectively. 

$8.1.15 Instrumentation does not alTect the 
motion of test devices during impact or rollover. 

$8.2 lateral moving barrier crash test condi- 
tions. The following conditions apply to the 
lateral moving barrier crash test : 

58.2.1 The moving barrier, including the im- 
pact surface, supporting structure, and carriage, 
weighs 4,000 pounds. 

58.2.2 The impact surface of the barrier is a 
vertical, rigid, flat rectangle, 78 inches wide and 
60 inches high, perpendicular to its direction of 
movement, with its lower edge horizontal and 
5 inches above the ground surface. 

58.2.3 During the entire impact sequence the 
barrier undergoes no significant amount of 
dynamic or static deformation, and absorbs no 
significant portion of the energy' resulting from 
the impact, except for energy that results in 
translational rebound movement of the barrier. 

58.2.4 During the entire impact sequence the 
barrier is guided so that it travels in a straight 
line, with no significant lateral, vertical or rota- 
tional movement. 

58.2.5 The concrete surface upon which the ve- 
hicle is tested is level, rigid and of uniform con- 
struction, with a skid number of 75 when meas- 
ured in accordance with American Society for 
Testing and Materials Method E-274— 65T at 40 
m.p.h., omitting water delivery as specified in 
paragraph 7.1 of that method. 

58.2.6 The tested vehicle’s brakes are disen- 
gaged and the transmission is in neutral. 

58. 2.7 The barrier and the test vehicle are 
positioned so that at impact — 

(a) The vehicle is at rest in its normal atti- 
tude; ‘ 

(b) The barrier is traveling in a direction 
perpendicular to the longitudinal axis of the ve- 
hicle at 20 m.p.h. ; and 

(R«v. May 19721 



(c) A vertical plane through the geometric 
center of the barrier impact surface and perpen- 
dicular to that surface passes through the driver’s 
seating reference point in the tested vehicle. 

S8.3 Rollover test conditions. The following 
conditions apply to the rollover test: 

58.3.1 The tested vehicle’s brakes are disen- 
gaged and the transmission is in neutral. 

58.3.2 The concrete surface on which the test 
is conducted is level, rigid, of uniform construc- 
tion, and of a sufficient size that the vehicle re- 
mains on it throughout the entire rollover cycle. 
It has a skid number of 75 when measured in 
accordance with American Society of Testing 
and Materials Method E-274-05T at 40 m.p.h. 
omitting water delivery as specified in paragraph 
7.1 of that method. 

58.3.3 The vehicle is placed on a device, 
similar to that illustrated in Figure 1, having a 
platform in the form of a flat, rigid plane at an 
angle of 23° from the horizontal. At the lower 
edge of the platform is an unyielding flange, per- 
pendicular to the platform with a height of 4 
inches and a length sufficient to hold in place the 
tires that rest against it. The intersection of the 
inner face of the flange with the upper face of 
the platform is 9 inches above the rollover sur- 
face. No other restraints are used to hold the 
vehicle in position during the deceleration of the 
platform and the departure of the vehicle. 

58.3.4 With the vehicle on the test platform, 
the test devices remain as nearly as possible in 
the posture specified in S8.1. 

58.3.5 Before the deceleration pulse, the plat- 
form is moving horizontally, and perpendicularly 
to the longitudinal axis of the vehicle, at a con- 
stant speed of 30 m.p.h. for a sufficient period of 
time for the vehicle to become motionless relative 
to the platform. 

58.3.6 The platform is decelerated from 30 to 
0 m.p.h. in a distance of not more than 3 feet, 
without change of direction and without 
transverse or rotational movement during the 
deceleration of the platform and the departure 
of the vehicle. The deceleration rate is at least 
20g for a minimum of 0.04 seconds. 



PART 571 ; S 208-9 



D-10 



Effective; 1/1/72; 8/15/73; 

•/15/75; 8/15/77 




[S9. Pressure vessels and explosive devices. 

S9.1 Pressure vessels. A pressure vessel that 
is continuously pressurized shall conform to the 
requirements of 49 CFR § 178.65-2, -0(b), -7, 
-9(a) and (b), and -10. It shall not leak or 
evidence visible distortion when tested in accord- 
ance with § 178.65-11 (a) and shall not fail in 
any of the ways enumerated in § 17S.65— 11 (b) 
when hydrostatically tested to destruction. It 
shall not crack when flattened in accordance with 
§ 178.65-I2(a) to the limit specified in § 178.65— 
12(a)(4). (37 F.R. 9222— May 6, 1972. Effec- 

tive: 6/2/72)] 

[S9.2 Explosive devices. An explosive device 
shall not exhibit any of the characteristics pro- 
hibited by 49 CFR § 173.51. All explosive ma- 
terial shall be enclosed in a structure that is 
capable of containing the explosive energy with- 
out sudden release of pressure except through 
overpressure relief devices or parts designed to 
release the pressure during actuation. (37 F.R. 
9222 — May 6, 1972. Effective: 6/2/72)] 



[Interpretation 

Several persons have raised questions as to 
what constitutes a “passive” restraint system — 
one that requires “no action by vehicle occu- 
pants” — as those concepts are used in Standard 
No. 208, Occupant Crash Protection (36 F.R. 
4600, March 10, 1971), effective January 1, 1972. 
Specifically, it has been asked whether occupant 
protection systems that require occupants to take 
protective action as a prerequisite to entering, 
seating themselves in, or operating a vehicle can 
qualify as a system that requires “no action.” 
One commonly discussed example of such “forced 
action” systems is a seatbelt interlock, which re- 
quires a seat belt to be fastened before the vehicle 
ignition system is operative. 

The concept of an occupant protection system 
that requires “no action by vehicle occupants” as 
used in Standard No. 208 is intended to designate 
a system that requires no action other than would 
be required if the protective system were not 
present in the vehicle. Under this interpretation 
the concept does not include “forced action” sys- 
tems as described above. 

This interpretation is not intended to rule out 
the possibility that further rulemaking action 
may be taken in the future to permit such systems 
in certain cases. (36 F.R. 8296 — May 4, 1971. 
Effective: 5/4/71)] 

36 F.R. 4600 
March 10, 1971 



(Rev. May 1972) 



PART 571 ; S 208-10 



D-ll 



| Docket No. 74-14; Notice 06) 

PART 571— FEDERAL MOTOR VEHICLE 
SAFETY STANDARDS 

Occupant Crash Protection 

This notice amends Standard No. 203, 
Occupant Crash Protection, to continue 
until August 31, 1977, the present three 
options available for occupant crash pro- 
tection in passenger cars. 

This extension of the present occupant 
crash protection options of Standard No. 
208 (49 CFR 571.208) was proposed 
July 19, 1976 (41 FR 29715), along with 
several other subjects that will be the 
subject of a future notice. Vehicle manu- 
facturers supported the proposal but re- 
quested that the options be extended 
indefinitely instead of being limited to 
a 1-year extension. Mr. Benjamin Red- 
mond advocated the use of an interlock 
system to increase usage of active belt 
systems. Ms. Lucie Kirylak expressed a 
preference for active occupant crash 
protection systems. The National Motor 
Vehicle Safety Advisory Council did not 
take a position on the proposal. 

Tire Secretary of Transportation has 
initiated a process for the establishment 
of future occupant crash protection re- 
quirements under Standard No. 208 (41 
FR 24070, June 14, 1976) . The Secretary's 
proposal addresses the long term Issues 
involved, and this 1-year extension of 
requirements is intended to provide the 
time necessary to reach that decision. 
Because a 1-year extension is consistent 
with the process that has been estab- 
lished and because a longer extension was 
not proposed for comment, the NHTSA 
declines to extend the existing require- 
ments as recommended by the manufac- 
turers. 

Other matters proposed in the notice 
that underlies this action will be treated 
at a later date, following the receipt of 
comments that are due on October 20, 
1976. 

The NHTSA notes that no effective 
date was proposed for the other matters 
addressed by the proposal. Those mat- 
ters Involve modification of the existing 
passive protection options so that they 
conform to the proposal of the Depart- 
ment of Transportation, and to reduce 
somewhat the femur force requirement. 
Also, further specification of dummy 
positioning in the vehicle was addressed. 
The agency proposes an immediate effec- 
tive date for these changes, because they 
represent relaxation of the requirements. 
However, the views of interested persons, 
particularly Volkswagen (which is cer- 
tifying compliance under one passive 
option) , are solicited by October 20, 1976. 

In consideration of the foregoing, the 
heading and text of S4.1.2 of Standard 
No. 208 (49 CFR 571.200) are amended 



bv chancing the date “August 31, 1976” 
to “August 31. 1977” wherever it appears. 

Effective date: August 26, 1976. 

(Secs. 103, 119. Pub. L. 89-5G3. 80 Stat. 718 (15 
U.S C. 1302, 1407); delegation o I authority 
at 49 CFR 1.50.) 

Issued on August 26, 1976. 

John V/. Snow. 

Administrator. 

I PR Dor. 73-25428 Filed e-26-7G;l:15 pm| 



FEDERAL REGISTER, VOL. 41, NO. 169 — MONDAY, AUGUST 30, 1976 

D-12 



APPENDIX E 

DISCUSSION OF 
STATISTICAL METHODS 



E-l 



B.l INTRODUCTION 



A number of statistical techniques can be considered as analytical tools 
to evaluate the effects of implementing FMVSS208. Four of these techniques 
are discussed in this appendix. 

• Regression Analysis 

• Contingency Table Analysis 

® Log Linear Analysis 

• Index Method Analysis. 

E, 2 REGRESSION ANALYSIS 

Statistics uses the term regression in two senses, one a broad sense and 
the other a restriction of the broad sense to a more "specific" one. Before 
we discuss these two (or more) concepts a word should be said about the term 
"regression" since it has various connotations that are not appropriate to most 
work. In the previous century, the British scientist, Galton, studied the "in- 
telligence" of fathers and first bom sons and found that if the father was 
more "intelligent" than average, the son usually was also, but he tended to be 
more average than the father. Galton referred to this phenomenon as "regres- 
sion of mediocrity." The first part of the term has stuck as the name of the 
whole technique of which Galton' s work is merely an early example. By the way, 
the above does not imply that the next generation is less intelligent than the 
previous, since, for example, for sons more "intelligent" than average, the 
fathers tend to be more average than the sons. 

In the current broad-sense usage, regression is the study of the func- 
tional relationship between a dependent variable and one or more independent 
variables. The choice of terms does not imply a cause- and- ef feet relationship. 
In fact, taking the extreme case, the dependent variable could be the cause and 
the independent variable the effect, e,g., if one tried to regress the 
size of a bomb on the amount of damage caused. 

It would be somewhat more precise to say that regression is the study of 
the mean or average structure of the dependent variable by means of the inde- 
pendent variates. One is usually not trying (in a primary sense) to find the 
variability of distribution of the dependent variable from the other variates. 
It is true that the research does look at the variability, but only in the 
second sense of wanting to see the stability or precision of the functional 
relationship of the average values of the dependent and independent variables. 



E-2 



Some examples of general regression would be: 

(1) Finding the relationship between a student’s college record 

(quantity point ratio) and his/her high school record, college 
boards and other records. 

(2) The position of a stellar object as a function of time and 

previous positions. 

(3) The probability of rain as a function of air pressure, previous 

weather, temperature, etc. 

(4) The probability of a person’s having blond hair as a function of 

whether or not he is Swedish, whether he is under 10 years, 
between 10 and 20, and over 20, etc. 

This general restricted concept of regression considers dependent varia- 
bles that have an interval scale, usually independent variables that are inter- 
val scaled, and a random error term. The random error term is assumed to be 
normally distributed. The independent variables are either values that can be 
adjusted by the researcher (e.g., the speed at which a test vehicle is driven) 
or normal random variables (e.g., the speeds of the cars in the population of 
cars considered is assumed to have a normal distribution). Both of these assump- 
tions imply, in the linear case, that the dependent variable is normally dis- 
tributed . 

As an example, we might be interested in a model regressing fuel consump- 
tion per mile F, on velocity of the vehicle V, the weight W, and the horsepower 
H. As a first approximation, we would have: 

F = y + aV + 3W + 6H + e, 

where e is the random error term. Since each of the independent variables ap- 
pears as a linear (first degree) term, we call this a linear equation. If we 
run the experiment under lab conditions and choose the speed, weight and horse- 
power values, these are considered fixed values and e is usually assumed to 
have a normal distribution. On the other hand, if the data are sampled (col- 
lected) from a random selection of actual vehicles, then the values of the in- 
dependent variables are not selected by the researcher and, in fact, have ran- 
dom distributions due to the random selection. However, the estimation of the 
usually unknown coefficients is, in both cases, carried out by least squares 
analysis. To accomplish this for all the data, we choose the values of m, a, 
b, c to minimize the summation 



E-3 



I (F i -m-aV i -bW i -cH i ) 2 t 

The objective is to find the precise equation that is closest to the ob- 
served data. If we consider the equation, F ® y + dV, then graphically we can 
obtain the following illustration. 




If the dots represent the data points, the line F » m + dV is chosen so that 
the sum of the squared distances represented by ")" is as small as possible. 
In order to judge whether or not the line gives a good fit to that data, we 
compare the original variability of the data from a horizontal line, 




F (average of F) 



with the sum of the squared distances from the sloping line. If the sloping 
line is a good fit there should be a substantial denumeration of the vari- 
ability. 



E-4 



In practice there are various difficulties that can only be handled 
approximately at this stage of statistical development. In general, data are 
not normally distributed. In many cases the linear equation does not fit the 
data well enough and higher order terms are needed. However, if V is normally 
distributed, then V^, V^, etc. are not. Nonetheless, the procedure seems to 
work quite well even when the assumptions of normality are not satisfied. One 
of its great advantages is its widespread use in many applied fields. Further- 
more, the procedures are quite standard and secondary analyses, such as comparing 
coefficients, can be done with little difficulty. On the other hand if the 
data, especially the dependent variable, are ordinal or nominal and if the 
range of the dependent variable is boun'ded, the results can be less than sat- 
isfactory. Also, if the dependent variable is not approximately normally dis- 
tributed, the procedure is not as efficient as others that use any distribu- 
tional knowledge. In addition, various statistical tests can be misleading if 
the distributional model does not reflect the true nature of the data in cer- 
tain aspects. 



E . 3 CONTINGENCY T ABL E ANALYSIS ' 

A more recent development has been that of contingency table analysis based 
on log linear models. While the basic contingency table analysis goes back to 
Karl Pearson and his chi-square test, the log linear means structure is a more 
recent development. 

In the Pearson chi-square v x c table, we usually have two factors or vari- 
ables, for example, degree of injury and speed. These are made categorical 
e.g., injury is on the scale of slight or - none, moderate or severe, while 
speed might be slow or fast. The body of the table contains the number of 
cases in each r and their respective probabilities (the latter) usually unknown 
iii practice category. 



INJURY 


SPI 


:ed 


Slow 


Fast 


Slight 
or None 


' 00 pn 


H0pi2 


Moderate 
or Severe 


50p 21 


80pp2 



150p + .j , 190p +2 



i 



210 



P1 + 



130 P2 + 



340 



'1 + 
and P 



P ll +P 12’ P + 1 = P 1 1 +P 21 ’ 



1 1 +P 12 + 



P 21 + P 22 



= 1. 



etc. 



E-5 



"k 

The usual chi-square analysis would give 

2 (100-92. 65) 2 , (110-117. 35) 2 , (80-72. 65) 2 , (50-57. 35) 2 „ ,, 

X 92765 + 117735 + 7 2765 57735 = 2 ' 44 

with 1 degree of freedom. The value 2.44 is not significant at a = 0.10. 

This result indicates that there is no dependence between speed and injury 

(for these data) and so the apparent discrepancies are due to random fluctuation. 

However, an interpretation of the effects of speed and injury is not all that clear. 



E .4 LOG LINEAR ANALYSIS 

A log linear model can be formulated such that 

log P.. = y + A. + M. 4 (AM) . . , 
ij i J ij 



where 



A x + A 2 = 0; M 1 + M 2 - 0; (AM)^ 4 (AM) 2 = 0; (AM) ±1 + (AM) 12 = 0; 

and A is the effect of injury (deviation of frequency of injury from the average) 
and M is the* speed effect and (AM) is the interaction, i.e., how much different 
speeds affect different levels of injury. This formula also gives the expected 
number E. . in each cell ij as 



log E. . = log NP. . = log N 4 log P. . 
p ij ij ij 



= log N 4 y 4 A. 4 M. + (AM) . . 

i 3 ij 

= y ' 4 A. 4 M. 4 (AM) . . 

il U 

where N is the total number of cases. 

The above x test tells us that (AM)_ = 0 for all vehicle speeds, A__ . 
Thus, we can say by appropriate analysis that the estimates of the E^ are E^ 

- 92.65, E 12 - 117.35, E = 57.35, and = 72.65 and y = 4.41, A = -A = 
0.237, M. = -M^ = -0.121. One can check these values of y, the M's and the A's 
given the appropriate E_'s. Uliile this analysis can be done without the log 
linear model for this simple case, the model can easily be extended to more 
variables with the interpretation being similar to the usual analysis of vari- 
ance. By extending the model we could include other factors such as weight 
of vehicle. 



2 

In general , x “ £ 



(Observed.. - Expected.. ) 

U y. 



Expected 



ij 



E-6 



An important property of the model is that it uses the discrete, multino- 
mial character of the data, something the normal model fails to do. This fact 
should make the analysis more precise. However, one failing of such an anal- 
ysis is that the dependent and independent variables are made discrete, which 
means that we cannot force the model to accept any ordering that we wish, e.g., 
we cannot force the effect of speed to be monotonic increasing. 

Another choice of analysis is to allow the contingency table analysis to 
have a functional relationship that has continuous and discrete independent 
variables. One would still have the advantage of the underlying multinomial 
distribution but this would allow the type of interval variables that are 
found in the regression concept. Namely, consider models of the form log P = 

u + A. + aC where A. is discrete as before and the C is a continuous variable. 

1 1 

Such an analysis should also consider interaction terms, e.g., what is the ef- 
fect of impact angle with or without a head restraint. 



APPENDIX F 

INTRODUCTION DATES FOR 
SIDE DOOR REINFORCEMENT 
BEAMS 



APPENDIX F : INTRODUCTION DATES OF SIDE DOOR REINFORCEMENT BEAMS 



Make 


Line 


Series 


Model 

Year 




AMC 


Javel in 


SST 


1971 






Basic 


1971 






AMX 


1971 


GM 








Buick 


Buick 


Electra 


1969 






La Sabre 


1969 






Riviera 


1971 




Special/Skylark 


Skylark 


1970 






GS 


1970 


Cadillac 


Cadillac 


Calais 


1969 






De Vi lie 


1969 






El Dorado 


1971 






Fleetwood El Dorado 


1971 






Fleetv/ood Brougham 


1969 






Fleetwood Seventy-five 


1969 






Fleetwood Sixty Special 


1969 


Chevrolet 


Chevelle 


Concours 


1970 






Mai ibu 


1970 






Nomad 


1970 






Greenbriar 


1970 




Chevrolet 


Bel Air 


1969 






Biscayne 


1969 






Caprice 


1969 






Kingswood 


1969 




Monte Carlo 


Monte Carlo 


1970 




Vega 


Vega 


1971 


Oldsmobile 


F-85/Cutlass 


F-85 


1970 




Oldsmobile 


Delta 88 


1969 






98 


1969 




Toronado 


Toronado 


1971 


Pontiac 


Firebird 


Firebird 


1970 






Esprit 


1970 






Formula 


1970 






Trans-Am 


1970 




Pontiac 


Bonneville 


1969 






Catal ina 


1969 






Executive 


1969 






Grand Prix 


1969 




Tempest/LeMans 


Le Mans 


1970 


CHRYLSER 








Dodge 


Challenger 


Challenger 


1970 






Challenger RT 


1971 


FORD 








Ford 


Fairline/Torino 


Gran Torino 


1972 




Ford 


Custom 


1971 






Galaxie 


1971 






LTD Brougham 


1971 




Mustang 


Mustang 


1971 






Grande 


1971 




Pinto 


Pinto 


1971 




Thunderbi rd 


Thunderbi rd 


1972 


Lincoln 


Lincoln 


Continental 


1971 






Continental Mark III & IV 


1971 


Mercury 


Cougar 


Cougar 


1971 






Cougar XR 7 


1971 




Mercury 


Marquis 


1971 






Marquis Brougham 


1971 






Monterey 


1971 




Montego 


Montego 


1972 






Montego MX, Brougham, & GT 


1972 



F-2 



APPENDIX G 

COST ESTIMATING 
METHODOLOGIES USED BY 
BLS , GAO, AND NHTSA 



G-l 



APPENDIX G: COST ESTIMATING METHODOLOGIES USED BY BLS , GAO, AND NHTSA* 

There are several methods for estimating the cost of consumers of FMVSS; 

GAO, NHTSA, and BLS use different methods. The methods used by the three or- 
ganizations were reviewed by reading descriptions of the methods and discus- 
sing the cost estimating problem with responsible individuals in each organ- 
ization. The main problems recognized by the three organizations were: 

• Magnitude of the problem . A large number of models must be 

examined according to size, popularity and different design concepts. 

• Lifetime Cost . The cost of any system includes initial costs 

and any added maintenance and operating costs. In standards 
where design changes lead to weight increases, the increased 
operating cost is an important consideration. 

• Innovation . The initial cost of a safety related feature may de- 

crease with engineering innovations. There is also the fact that 
substitution of lighter materials can reduce the additional operat- 
ing costs. 

• Marginal Cost Concept . If all safety standards were eliminated, 

manufacturers would not remove all safety features from all 
vehicles. For some vehicles in certain vehicle classes, some 
safety features would remain. The likely response of the manu- 
facturers would be to design safety features in line with the 
overall design and price aspects. Therefore, the argument was 
made that the consumer cost of safety features is the difference 
in cost between what the manufacturers would provide if there were 
no standards versus what they do provide to meet the standards. 

With regard to the last point, there are conceptual and practical difficul- 
ties. One problem arises in the comparison of the marginal cost of the safety 
feature with the estimated benefit due to injury reduction. The relative com- 
parison would be more difficult and perhaps less meaningful if one had to es- 
timate hypothetical costs and benefits which would exist without the standard. 

The practical problem would be trying to estimate the hypothetical cost and 
benefit levels. To estimate a meaningful measure imprecisely is no worse than 
to estimate a meaningless measure precisely. 

In general, the cost estimating methods vary in proportion to their reli- 
ance on industry-supplied estimates. The GAO relied very heavily on the 



* 

BLS - Bureau of Labor Statistics 
GAO - General Accounting Office 

NHTSA - National Highway Traffic Safety Administration 



G-2 



manufacturers’ cooperation and help in providing documentation to develop the 
estimates which they used in their report.* The GAO has a unique position 
within the Federal Government and had strict agreements on the confidentiality 
of information with the manufacturers. NHTSA has the authority to request 

similar information from the manufacturers. However, NHTSA relies on its own 
inhouse engineering staff to provide initial estimates of the costs. Discus- 
s ion continues with the manufacturers on the accuracy of these initial esti- 
mates. Written interrogatories may take place to improve the initial NHTSA 
estimates. An additional step which is taken if the estimated costs are still 
uncertain is to contract with independent industrial engineering firms for 
estimates. The BLS gathers information on the cost of safety features in or- 
der to exclude those price changes from the compilation of the Consumer Price 
Index (CPI) for new cars. The addition of the safety features is classified 
as a quality change in the product. The manufacturers reply to written ques- 
tionnaires on the cost of specific safety related equipment on a selected sam- 
ple of 16 vehicles. The GAO and BLS estimates are only initial costs. NHTSA 
tries to estimate the additional costs over the life of the safety feature; 
initial, operating, and maintenance. The individual cost estimating methods 
are described in more detail below: 

NHTSA’ s Approach ** 

Motor Vehicle Programs has established a three-step effort for estimating 
costs in support of rulemaking. The first effort is inhouse cost estimates by 
means of a standardized high-volume industrial-processing building-block cost 
estimating methodology; the second is dialog and questionnaires with industry, 
out cf which information and estimates would be provided. The third effort is 
contracting with non-government sources to establish neutral cost estimates. 
The inhouse cost estimating methodology covers 1) direct manufacturing; 2), 
indirect manufacturing; 3) capital investments; 4) manufacturing markups; 

5) dealer markups; 6) taxes; 7) lifetime operating; 8) lifetime main- 
tenance life-cycle cost factors. 



Report to the Committee on Commerce, United States Senate by the Comptroller 
General, Effectiveness 3 Benefits 3 and Costs of Federal Safety Standards for 
Protection of Passenger Car Occupants. Government Accounting Office, CED-76-121. 
^uly 7, 1976. 

Taken from NHTSA’ s Approach for Determining the Consumer Cost of Motor Vehicle 
Programs’ Rulemaking Programs by Charles Westphal, Jr. of NHTSA’ s Motor Vehicle 
Program^ Engineering Systems Staff and discussions with the author. 



G-3 



The basic precept of a building-block cost estimating methodology is that 
estimates, representative of the average impact, can be generated by utilizing 
historical information instead of relying on traditional methods for estimating 
costs. Fractional methods depend on the availability of 1) well-defined descrip- 
tions of proposed design concepts; 2) detailed bills of material; 3) manufac- 
turers’ detailed processing operations; 4) personnel with specialized judgment; 
and 5) individual manufacturer product plans. 

In contrast to the traditional methods, "the estimating methodology 1) is 
useable with minimum aid from cost specialists, 2) considers the relationship 
of the timing of the effective date of the safety standard relative to the pro- 
duction cycle, 3) generates estimates credible to motor-vehicle cost and fin- 
ancial professionals, and 4) provides reasonable confidence backup data for 
public, congress, industry and court validation that motor-vehicle programs 
are in the public interest; i.e., reasonable, practicable and appropriate by 
Public Law 89-563."* 

The principal steps involved in generating cost assessments via NHTSA's 
high-volume industrial-processing building-block methodology are 1) safety per- 
formance requirements are transformed into representative design concepts, 

2) design concepts are broken down into representative high-volume industrial- 
processing operations, materials, and/or labor quantities and then 3) the costs 
of design concepts are determined by applying the cost per pound experienced in 
similar high-volume industrial-processing operations to the number of pounds 
of material making up the design. 

In addition to the initial cost to the consumer, lifetime maintenance and 
operating costs are also calculated. These latter are important in comparing 
alternative design concepts. Cooperation has increased between the manufactu- 
rers and NHTSA which has decreased the time and cost of estimating the cost of 
safety standards from the level required by the more traditional method. How- 
ever, it has been reported that considerable effort is still required to obtain 
and review the relevant detailed design drawings, identify the individual 
design concepts, and estimate the manufacturing costs. 



* 

Ibid. 



G-4 



Government Accounting Office (GAO) Approach 

The GAO approach was paraphrased above as the traditional method, obtaining: 

1. Well-defined descriptions of proposed design concepts. 

2. Detailed bills of material. 

3. Descriptions of manufacturers' detailed processing operations. 

4. Personnel with specialized knowledge. 

5. Individual manufacturer product plans. 

Although conversations with GAO staff were very helpful, they were con- 
strained by professional and proprietary factors. The GAO had considerable 
cooperation from the three largest domestic auto manufacturers and some in- 
formation from two foreign manufacturers. Given the large number of standards 
considered and the alternative safety features designed, the GAO studied only 
selected representative models for each Standard. A significant factor is 
the different cost accounting systems used by each manufacturer.** 

Bureau of Labor Statistics (BLS) Approach ^ 

The BLS collects information on the costs of safety features so that these 
and other "quality" changes in the price of a new automobile are not reflected 
as an inflationary price increase. The BLS has only 16 cars in their sample 
and weights the factors for an average figure. BLS reports these quality 
changes for safety standards only after their effective date. If some manufac- 
turers introduce safety features before the effective date, that change is ini- 
tially reflected in an "other" quality change category. BLS only reports the 
industry supplied estimate of safety feature price changes once so that subse- 

■ 4 * 4 * 

quent reductions in the cost of that feature are not taken into account. The 
BLS estimates of the cost of safety features may be good for the year in which 
thev are introduced or upgraded and could serve as a check on estimates by GAO 
and NHTSA. 



Mr. Don Cluff, Project Manager, and Mr. John Pennington, Audit Supervisor 
discussed the basic approach and problems with CEM staff on Nov. 5, 1976, 

** 

The accounting system may be related to the degree of vertical integration 
enjoyed by the firm, 

^Personal communication with M. Voorhees, CPI Commodity Specialist, Bureau of 
Labor Statistics, Nov. 9, 1976. 

^This fact will have a cumulative effect and thus the CPI for new cars might 
underestimate the real price inflation significantly. 



APPENDIX H 



A STATISTICAL METHOD FOR 
COST DATA ACQUISITION: 
HOW TO SELECT THE MAKE 
AND MODEL PRODUCED BY 
A MANUFACTURER 



H-l 



APPENDIX H ; STATISTICAL DISCUSSION ON CHOOSING A PARTICULAR MAKE/MODEL WITHIN 
MANUFACTURER FOR COST DATA ACQUISITION 

Consider any cell in the experimental design corresponding to a particular 
manufacturer and market class. Suppose within this cell there are K different 
possible cars to choose with known sales volumes n^, n^ , . ..n^ (let n = ^ n^) . 
Suppose also the respective unknown costs are c^, c^...^* i = l 

We seek an estimate of the overall average cost 

n 

T = I c, -i 



based on one observation. 



Any decision rule may be described by a set p 1 ,...p where p is the 

tli -L Ic i 

probability of selecting the i possible car and then obtaining its cost c^. 

The risk associated with any rule, under squared error loss (obviously 
appropriate under variance considerations) is 



l (c, - c ) 2 p. 



The natural inclination at this point is to attempt to minimize this risk 
over the p_^. The answer is set p, = 1 at c^ closest to c. But this is clearly 
worthless since the c^ are unknown. (If they were known, c would also be known 
and there would be no problem.) 

Hence, the choice of the p^'s can only depend on the n^. The natural 
approach suggests the unbiased estimator p^ = _^i so that the expected value 
of the estimator is c. The associated risk n is 



I (c, - c) 



2 n i 



We wish to examine which of these is the smaller. First we solve the 
problem if k=2 in which case n^/n > 1/2. 

Claim: (c x - c) 2 < (Ci - c) 2 2^- + (c, - ^) 2 ^ 



1 2 



1 2 



n l n 2 

Proof: Obvious: plug in c = c, ; + c„ : and verify. 

1 n 1 +n 2 2 n 1 +n 2 



H-2 



More generally, 



if we write 



n. 



7 = c 



1 n 



n„ 



k 

l n i c i 
i=2 1 1 

n 



= c 



+ c» 



ti-n, 



n 



1 n 



k 

where c 1 = V 

i=2 n-n i 



In other words, c is the weighted average of with the weighted 



average of the remaining c^s. Then, 



£ (c r c>2 it = (c i - C)Z it + (c r c ' + it 



_ N 2 “l 



2 i 



i=l 



i=2 



n . 



compared with 



r. n« n (n-n-, ) k ~ . 

- <C 1 " c)2 ~t + (c'-r) 2 - T ^ + fci-c') - > 



(c x - c) 



n 



But if — 
n 


> 1/2 


then c 


1 • G « j 


r-H 

O 


c) 2 


< (c 


or 


(C 1 ' 


c) 2 


(n-n 


n 


or 


(c l " 


IT) 2 


< (c 



2 



2 



n 



„ n 

—\2 1 



2 ( n-n ^) 



H-3 



Since the circled term is >0, selecting c^ via n^ clearly gives the smaller 

risk. If < 1 , there is no "best" solution. The better choice can only be 

n 2 n 1 

made knowing the c . If __1 is close to -z, the circled term should still 

1 n 

be large enough to make selecting c^ via n^ the better choice. 

On the other hand, if all the n_^ are about the same, i.e., 



n i ^ — then 

— k 
n 



c ~ 



Ic 



, r (c.-c) 2 
and — i 
i 



n . 

l 



(c.-c) 2 



— . 2 



i.e. , the "average" (c.-c) 



— 2 , 

is no better than any particular (c^-c) Hence, 



again selecting c^ via should still be as effective as randomizing. 






H-4 



APPENDIX I 

FLOWCHARTS OF 
SELECTED ANALYSES 



i-l 




Figure 1-1. Proposed Statistical Analysis Scheme for Evaluating 
FMVSS 215 (Exterior Protection). 



1-2 



Step 1 



Step 2 



Step 3 




Questions to be Answered 

Is fuel system rupture less frequent 
in post-standard, towaway s t where 
no obvious aging ef feats are 
evident? 



Are there identifiable discontinuities 
and/or ohanges of slope in the trend 
of occurrence of obvious aging 
effects of fuel system components 
vs. car age? 



Are there identifiable discontinuities 
and/or changes of slope in the trend 
of rupture in towaway cases 3 where 
there are obvious aging effects in 
fuel system components vs. car age? 



Figure 1-2. 



Proposed analysis scheme for evaluating FMVSS 301 
with fuel system rupture towaway data. 



1-3 




Figure 1-3. Proposed statistical analysis scheme for evaluating 
FMVSS 301 with vehicle fire and fuel spillage data. 



1-4 




Figure 1-4. Proposed Statistical Analysis Scheme for analyzing 
fire-related fatal automobile accidents. 



1-5 




< 

> 



o 

70 



70 



C= IE 

> 

H 
—< TJ 
O v. 



ro 

-Cr 
to ro 
o 



CD — 



1