|NHTSA Report Number DOT HS 807 489||November 1989|
Charles J. Kahane, Ph.D.
Federal Motor Vehicle Safety Standard 206 - Door Locks and Door Retention Components - is aimed at reducing the likelihood of occupant ejection in crashes. The industry steadily improved door lock design during the 1960's. Standard 216 Roof Crush Resistance - is designed to reduce deaths and injuries due to the crushing of the roof into the passenger compartment in rollover crashes. Hardtops were redesigned as pillared cars, with stronger roof support. evaluation analyzes the effectiveness and benefits of stronger door locks and roof structures in rollover crashes of passenger cars. It also estimates the cumulative effect on fatality risk - for unrestrained occupants in rollover crashes - of all safety standards and vehicle modifications of the 1963-82 era. The study is based on statistical analyses of FARS, Texas, NCSS, NASS and MDAI data and roof crush tests on pre- and post-Standard 216 cars. It was found that:
Executive Order 12291 (February 1981) requires agencies to evaluate their existing regulations. The objectives of an evaluation are to determine the actual benefits - lives saved, injuries prevented, damage avoided - and costs of safety equipment installed in production vehicles in connection with a standard.
The goal of this report is to evaluate the life saving benefits associated with Federal Motor Vehicle Safety Standards 206 and 216 for unrestrained occupants of passenger cars. Standard 206 - Door Locks and Door Retention Components - took effect on January 1, 1968 and is aimed at "minimizing, the likelihood of occupants being thrown from the vehicle as a result of impact." Standard 216 - Roof Crush Resistance - has applied to passenger cars since September 1 , 1973 and its purpose "is to reduce deaths and injuries due to the crushing of the roof into the passenger compartment in rollover accidents." Vehicle modifications in response to these standards have been piecemeal and gradual. The domestic auto industry anticipated Standard 206 by many years and had been making incremental year to year improvements in door design throughout 1956-68. Standard 216 took effect in the middle of the gradual change in roof styling from true hardtops to pillared hardtops, a process which stretched over most of the 1970's (and may have been motivated by other factors in addition to Standard 216).
It is best to study Standards 206 and 216 in the context of the overall trend in fatality risk of unrestrained occupants of passenger cars of model years 1963-82 in rollover crashes, for this is the type of crash in which strong roofs and better door locks are especially likely to have benefits. Standards 206 and 216, however, are not the only vehicle factors which affected fatality risk in rollover crashes during the 1963-82 period. A major task of the evaluation is to study the overall fatality trend and identify what changes are due to improved door locks and roof crush strength, as opposed to other vehicle factors.
Rollover crashes are a major safety problem, resulting in about 4,000 fatalities a year to occupants of passenger cars. A noteworthy aspect of rollovers is that many of the fatal crashes do not involve great amounts of force or destruction to the car. Two thirds of the fatalities in rollovers involve occupants being ejected from the car, often in crashes with low damage.
A number of strategies are available to reduce deaths and injuries in rollovers. The best single measure is to use safety belts. Recent studies have shown that belts are exceptionally effective in rollovers, reducing fatality risk by 70 percent or more. Many occupants do not use manual safety belts, however, especially those who are likely to become involved in severe rollovers.
A first line of defense against rollover fatalities is to prevent a car from rolling over. The next line of defense is to keep the occupant inside the car. As noted above. many of the ejections occur in crashes of low severity. The design of doors and their locks, latches and hinges is crucial here; so is the retention and integrity of windows. Next, the occupants' living space within the passenger compartment must be maintained. The roof has to be strong enough to resist severe compression when the car rolls over. Finally, impacts with the interior surfaces of the passenger compartment should not injure the occupants.
The principal analysis technique of the evaluation is to define and compute year to year trend lines or risk indices: e.g., an overall fatality -risk index, a crashworthiness index and a roof crush strength index.
The first set of trend lines generated in the evaluation is shown in Figure 1 . The curve connecting the U's on the figure is the rollover propensity index for passenger cars by model year. It is based on Texas accident data; rollover propensity is the ratio of rollovers to frontal impacts with fixed objects, with some adjustments. This measure of "rollover propensity" combines the concepts of directional stability (tendency of cars to stay under the driver's control and on the road) and rollover stability (tendency of cars to remain upright, given exposure to off-road tripping mechanisms). The rollover propensity index starts at a level close to 85 in model year 1963 and briefly rises to the 90's before dropping to a low of about 80 by 1970. After model year 1970, rollover propensity rises steadily year after year to an all time high close to 120 in model year 1982. It is well known from the literature that rollover propensity is highly correlated with car size Parameters such as track width, wheelbase, curb weight, or the height of the center of gravity, although it is not clear which one of those intercorrelated parameters is more influential than the others. Obviously, the steady increase in rollover proneness after 1970 coincides with the trend to vehicle downsizing and the shift from wide, long and heavy domestic cars to narrower, shorter and lighter imports and subcompacts.
Figure 1: ROLLOVER PROPENSITY INDEX BY MODEL YEAR
(1975-80 average = 100)
The evaluation is not an investigation of the "causes" of rollover. Nevertheless, the approach used to calculate the benefits of vehicle modifications necessitates checking if there are any important factors besides car size significantly correlated with rollover proneness . A statistical analysis of the Texas data shows that much of the variation across makes. models and model years can be explained by car size parameters such as track width, wheelbase and curb weight - with one important exception: the pre-1969 Volkswagen Beetle had a rollover rate even beyond what would be expected from its narrow, short and light design. The curve connecting the A's in Figure I is the rollover propensity index after adjustment for year to year changes in track width, curb weight and wheelbase. It starts at 110 and rises in the mid 1960's as the Volkswagen Beetle became more popular. During 1967-69, following important changes in the suspension and wheels of Volkswagen Beetles, the index drops quickly to 100 and it has remained essentially unchanged since 1970. There may have been other models with exceptional rollover rates, but none of them had sufficiently high sales or extreme rollover rates to pull the index (average for all cars) away from 100. Rollover propensity, on the average, has become very well correlated with car size.
The curve connecting the U's in Figure 2 is the most comprehensive measure of vehicle performance in this evaluation. It is the overall rollover fatality risk index for passenger cars by model year, comprising the net effects of changes in rollover propensity and crashworthiness. it is based on Fatal Accident Reporting System data; fatality risk is the ratio of fatalities in rollovers to those in frontal impacts with fixed objects, with some adjustments. The fatality risk index starts at about 107 in model year 1963 and drops quickly at first, then more slowly to a low in the upper 80's by model year 1973. After model year 1975. the fatality index rises at an increasing rate year after year to an all time high of about 123 in model year 1982. In other words, an occupant of a 1982 car has 15 percent higher likelihood (123/107) of dying in a rollover crash than a 1963 car occupant, under similar driving conditions.
The principal reason that newer (smaller) cars have higher rollover fatality risk is that they have higher rollover propensity: the more rollovers, the more deaths. A major task of the evaluation is to separate out the effects of changes in crashworthiness from changes in rollover propensity. The curve connecting the A's in Figure 2 is the crashworthiness index for rollovers: rollover fatality risk adjusted for rollover propensity. Here, the results are more favorable for new cars. The crashworthiness index starts at just over 120 in model year 1963 and drops quickly at first. then more slowly till it reaches 100 in the early 1970's. It has been close to 100 since model year 1975.
FIGURE 2: ROLLOVER FATALITY RISK INDEX BY MODEL YEAR
(1975-80average = 100)
More detailed fatality indices make it possible to study the effects of individual vehicle modifications. Figure 3 is the crashworthiness index for ejection fatalities only, relevant to the analysis of door locks and Standard 206. Ejectees account for two thirds of the rollover fatalities. The ejection fatality risk index starts at about 125 in model year 1963 and drops sharply during the mid 1960's. when the manufacturers significantly improved door latches. It continued to drop at a slower rate during the late 1960's, as manufacturers implemented further improvements. (A small portion of the reduction may be due to adhesive bonding of the windshield, a vehicle modification associated with Standard 212.) The ejection index reached 100 in 1970-71 and has stayed close to 100 ever since.
FIGURE 3: ROLLOVER EJECTION FATALITY RISK INDEX BY MODEL YEAR
(Adjusted for rollover propensity: 1975-80 average = 100)
Figure 4 is the corresponding crashworthiness index for occupants who were killed without being ejected. In general this is a far more severe group of crashes, for close to 75 percent of ejectees would have survived if they had stayed in the car. The nonejection fatality index is close to 108 throughout the 1960's. During model years 1972-76, as true hardtops were changed to pillared hardtops, the index drops to 100 and it stays close to 100 thereafter. A separate analysis of fatality risk in hardtops and sedans confirms that the fatality reduction is due to changes from true to pillared hardtops.
FIGURE 4: ROLLOVER NONEJECTION FATALITY RISK INDEX BY MODEL YEAR
(Adjusted for rollover propensity: 1975-80 average = 100)
The roof crush strength of passenger cars was studied in laboratory tests and accident data Tho Standard 216 compliance test data base of 108 new, post-standard cars was supplemented by 20 tests of used cars, including 14 pre-standard vehicles. Figure 5 is an index of the average performance on the Standard 216 test by model year. The index is obtained by statistically transforming the actual inches of crush to a normal variable and adjusting for biases that happened because some of the test samples emphasized certain manufacturers or market classes. The crush depth index is zero for the average car, negative for a stronger than average roof, positive for weaker; the index values do not readily translate back to actual inches of crush. Cars of the mid 1960's actually had the strongest roofs on the tests, with a normalized average crush depth of -0.7. In the later 1960's, large cars emphasized a look with a wide, flat roof. That resulted in weaker roof crush performance, with a normalized crush depth of +0.9 in model year 1970. From model year 1974 onwards (post-Standard 216), roof crush resistance is better than in 1970 and the normalized score is usually close to 0 (average strength). A more detailed look at the laboratory test results shows that most cars easily exceeded the requirements of Standard 216, even before the standard took effect. About half the cars with marginal performance on the Standard 216 test were full-sized hardtops, although not all hardtops had that problem. The elimination of true hardtops during the 1970's helped eliminate many of the marginal performers.
FIGURE 5: ROOF CRUSH DEPTH IN STANDARD 216 TESTS BY MODEL YEAR
(Normalized and adjusted for market class and manufacturer)
The Standard 216 compliance test is only one way of measuring roof strength. Another is to look at the actual amounts of roof crush in rollover accidents. The extent of roof crush is documented in the Collision Deformation Classification by a scale ranging from I (minimal damage) to 9 (extreme damage), in data on the National Accident Sampling System, National Crash Severity Study and Multidisciplinary Accident Investigation files. After the data are corrected for reporting differences between the files and adjusted for car size, the average crush depth rating is graphed by model year in Figure 6. The curve connecting the S's indicates the trend in crush for sedans, pillared hardtops and other cars with full B pillars. Roof performance hardly changed during 1963-82 dropping from an average of 3.7 crush zones in cars of the mid 1960's to 3.6 by the early 1980's. The curve connecting the H's depicts the trend in crush for true hardtops. During the mid 1960's, they were about as strong as sedans. Throughout 1968-75, true hardtops had significantly weaker roofs -than pillared cars, with crush extending 4 zones on the average. The elimination of true hardtops in the 1970's helped Improve the overall average roof crush strength of cars. In summary, the analyses show that certain hardtop designs had weaker roofs and higher nonejection fatality rates in rollovers than other cars of the same size. The elimination of those designs saved lives.
The critical problem in developing safety indices for motor vehicles is separating the true effects of vehicle modifications from other factors that could bias the indices: changes in driving habits. changes in roadway or exposure patterns, year to year inconsistencies of definitions or reporting on accident data files. There are no algorithms for identifying and removing biases; it is up to the analyst to judge what is a bias and what is the best method to remove it. The validity of the indices in Figures 1-6 depends on these judgements. The accident and test data used in this evaluation contain generous samples for cars of the 1970's but thins out for the oldest and youngest cars. That made it impossible to study cars before 1963 or after 1982 and even for 1963-64 and 1981-82 the sampling errors are visibly larger than for the middle years.
FIGURE 6: ROOF CRUSH EXTENT IN ROLLOVER CRASHES BY MODEL YEAR
(Adjusted for car size)
One complication in the analyses is that vehicle size parameters such as track width, wheelbase and curb weight are highly intercorrelated - i.e., "large" cars tend to be wider, longer and heavier than "small" cars . While the statistical analyses used here accurately identify the increase in rollover propensity for the typical small car relative to the typical large car, they may err in estimating what portion of the increase is attributable to any one parameter. Specifically, it is inadvisable to use the formulas of this report to predict what might happen in the future if a single parameter (say, curb weight) is changed while others are held constant.
The evaluation of Standard 206 is limited to passenger cars in rollover crashes. The standard also applies to light trucks, vans and multipurpose vehicles and it is likely to have benefits in side impacts as well as relievers; however, those additional benefits could not be estimated by the approach used in this report.
The results in this report are based on a population of mostly unrestrained occupants. During the years of data covered in the report, belt usage in rollover crashes was too low to provide a sample adequate for the analysis of Standards 206 and 216 for belt users.
Despite the benefits associated with improved door locks and roof crush resistance, rollover crashes continue to account for a high percentage of fatalities in passenger cars, light trucks and utility vehicles. Thousands of occupant fatalities involve ejection through side windows or open doors. NHTSA has undertaken a comprehensive research and rulemaking program to find ways to reduce the number of rollover crashes and to protect occupants in those crashes. The agency is developing new accident data bases to improve understanding of the causes of actual rollover crashes. Mathematical and computer models are being developed to simulate vehicle dynamics and occupant kinematics in rollovers. Staged rollover crashes provide data for validating the simulation models and preliminary design of a "standard" rollover test facility. The agency is studying the strength of current door lock systems and developing glass plastic side windows designed to reduce the risk of occupant ejection in crashes. In September 1988, NHTSA granted a petition for rulemaking to establish a standard to protect against unreasonable risk of rollover. The proposed upgrade of the side impact protection standard includes a requirement that the doors remain closed during the impact test; the objective is to reduce the risk of occupant ejection through open doors.
The ultimate goal of the evaluation is to identify the individual vehicle modifications that affected fatality risk during the 1963-82 period and estimate the change in fatalities for each of them. Based on an examination of the trends in Figures 1-6 as well as more detailed analyses, the study's principal findings and conclusions on the individual vehicle changes are the following:
Side door performance
A number of significant improvements to door latches and locks of domestic and imported cars were implemented during 1963-68. They save an estimated 400 lives per year by preventing about 15 percent of the ejections in rollover crashes.
Cars with 2 doors have 28 percent higher ejection risk in rollovers than 4 door cars, even after adjusting for differences in car size and exposure patterns. The market shift from 43 percent 2 door cars in model year 1963 to 67 percent in 1974-75 resulted in an increase of 150 fatalities per year.
Conversely, the market shift from 67 percent 2 door cars in 1974-75 back to 45 percent 2 door cars by 1982 has saved 140 lives per year.
Roof crush resistance
True hardtops have approximately 15 percent higher risk of a nonejection fatality in a rollover crash than pillared cars of the same size and exposure pattern.
During the 1970's, true hardtops were restyled as pillared hardtops or sedans, saving an estimated 110 lives per year.
13 of 128 cars tested had "marginal" performance on Standard 216 (more, than 4 inches of roof crush at a force level 10 percent above the Standard 216 requirement). Six of these 13 cars were full-size hardtops.
Narrower, lighter, shorter cars have higher rollover rates than wide, heavy, long ones under the same crash conditions. During 1970-82, as the market shifted from large domestic cars to downsized, subcompact or imported cars, the fleet became more rollover prone. That may have been partly offset by increases in the track width of some imported cars after 1977. The net effect of all car IW changes since 1970 is an increase of approximately 1340 rollover fatalities per year.
Before 1969, the Volkswagen Beetle with the swing axle suspension had an even higher rollover rate than would be expected for a car of its size. Redesign of the suspension and wheels during model years 1967-69 brought the rollover rate down to the expected level, saving 280 lives per year.
The fatality or injury rate per 100 rollover crashes is not a valid measure of crashworthiness in comparisons of cars of different sizes. Cars that tend to roll over easily (small, narrow cars) do so in crashes of intrinsically low severity. These rollovers have low injury rates. Larger cars would not roll over at all in those circumstances; when they do roll over it's a severe crash likely to result in injuries. The fatality rate per 100 crashes is lower for small cars, even if they are no more crashworthy.
Summary of annual effects of vehicle modifications rollover fatalities
|Vehicle Modification||Date||Lives per Year|
|Improved door locks (Standard 206)||1963-69||400|
|Shift from 4 door to 2 door cars||1963-74||150|
|Adhesive bonding of the windshield||1963-82||40|
|Improved suspension for Volkswagen||1967-69||280|
|Shift to subcompact & imported cars||1970-82||1,220|
|Curtailed production of true hardtops (Standard 216)||1971-77||110|
|Downsizing of existing car lines||1975-82||350|
|Shift from 2 door back to 4 door cars||1976-82||140|
|Wider tracks for some imported cars||1977-81||230*|
|NET LIVES LOST PER YEAR||520|