Abstract

Rollover crashes kill more than ten thousand occupants of passenger vehicles each year.  As part of its mission to reduce fatalities and injuries, since model year 2001 NHTSA has included rollover information as part of its New Car Assessment Program (NCAP) ratings.  One of the primary means of assessing rollover risk is the static stability factor (SSF), a measurement of a vehicle’s resistance to rollover.  The higher the SSF, the lower the rollover risk.  This report tracks the trend in SSF over time, looking in particular at changes in various passenger vehicle types.

Data are presented for overall fleet average SSFs by vehicle type over a number of model years.  Passenger cars, as a group, have the highest average SSF, and these have remained high.  SUVs have substantially improved their SSF values over time, especially after model year 2000, whereas those of pickup trucks have remained consistent over the years.  Minivans showed considerable improvement since they were first introduced, while full-size vans showed a small but steady improvement.  In model year 2003, the sales-weighted average SSF was 1.41 for passenger cars, 1.17 for SUVs, 1.18 for pickup trucks, 1.24 for minivans, and 1.12 for full-size vans.

Introduction and Background

Rollovers are among the most severe traffic crashes, and are of particular concern for occupants of light trucks and vans (LTVs) - including pickup trucks, sport utility vehicles (SUVs), minivans and full-size vans up to 10,000 pounds Gross Vehicle Weight Rating (GVWR).  While only about three percent of all passenger vehicle (passenger car and LTV) crashes involve rollover, according to the 2003 Fatality Analysis Reporting System (FARS), one third of all passenger vehicle occupants that lost their lives were in vehicles that rolled over, a total of 10,376 rollover deaths.  Of these, 4,433 were in passenger cars, 2,639 in sport utility vehicles (SUVs), 2,569 in pickup trucks, and 724 in vans, and the remaining 11 in other or unknown types of light trucks.  Passenger cars had the lowest rollover fatality rate (23 % of fatalities were in vehicles that rolled over), while SUVs had the highest, 59 %.  Similarly, according to the General Estimates System (GES), six percent of passenger car occupants that were injured were in vehicles that rolled over.  LTV rates of rollover-related injured occupants were higher – nine percent of those injured in vans, thirteen percent in pickup trucks, and twenty percent in SUVs.  Looking at occupant fatalities per 100,000 registered vehicles, passenger cars had the lowest rate at 3.69, with vans similarly low at a rate of 3.83.  The rates for pickup trucks (7.18) and SUVs (10.22) were much higher.[1]  Clearly, rollover crashes are a major safety problem for all classes of light vehicles, particularly LTVs.

On June 1, 2000, the National Highway Traffic Safety Administration (NHTSA) proposed adding a measure of rollover resistance to the NCAP program, to begin in model year 2001.[2]  It was the agency’s belief that consumer information on the rollover risk of passenger cars and LTVs would influence some consumers to purchase vehicles with a lower rollover risk, and inspire manufacturers to produce vehicles with a lower rollover risk.  This, in turn, would reduce the number of injuries and fatalities from rollover.

Section 12 of The Transportation Recall Enhancement, Accountability, and Documentation (TREAD) Act[3], enacted November 1, 2000, directed the Secretary of the Department of Transportation to develop a rollover test for motor vehicles, to carry out a program of rollover tests, and to develop and disseminate consumer information on rollover performance.[4]  This responsibility was subsequently delegated to NHTSA.  As planned, in model year 2001 NHTSA began including rollover resistance information in its NCAP.

The original rollover resistance ratings were determined solely from the vehicle’s static stability factor (SSF).  The SSF of a vehicle is an at-rest calculation of its rollover resistance, based on its most important geometric properties. Basically, SSF is a measure of how top-heavy a vehicle is.  A vehicle's SSF is calculated using the formula:

where

T=track width

H=height of the center of gravity of the vehicle

The track width is the distance between the centers of the right and left tires along the axle. The location of the center of gravity is measured in a laboratory to determine the average height above the ground of the vehicle's mass.  The lower the SSF number, the more likely the vehicle is to roll over in a tripped single-vehicle crash.  A higher SSF value equates to a more stable, less top-heavy vehicle.  SSF values across all vehicle types typically range from around 1.00 to 1.50. Most passenger cars have values in the 1.30 to 1.50 range.  Higher-riding SUVs, pick-up trucks, and vans usually have values in the 1.00 to 1.30 range.[5]

Throughout the development of testing procedures, both linear and logistic regressions were run to determine and verify the relationship between SSF and rollover.[6]^,[7],[8]  It was found that the correlation of SSF to rollovers in single vehicle crashes is exceptionally robust in an area as complex as rollover.

Commenters on the Federal Register announcement had suggested that the rating system did not go far enough, and suggested that a dynamic test be added.[9]  NHTSA sought to expand the information provided in its rollover resistance ratings, and requested comments on the subject of dynamic rollover testing.[10]  After evaluating numerous driving maneuver tests for the dynamic rollover consumer information, NHTSA published its findings in a notice of proposed rulemaking in the Federal Register.[11]  The final modifications to the rollover resistance ratings in NCAP, including the addition of dynamic rollover tests, were announced[12] and took effect beginning with model year 2004 vehicles.

The new dynamic maneuvering test used by NHTSA to help evaluate rollover risk utilizes a heavily loaded vehicle (to represent a five-occupant load), and a full tank of gas.  Using a fishhook pattern, the vehicle simulates a high-speed collision avoidance maneuver—steering sharply in one direction, then sharply in the other direction—within about one second.  Test instruments on the vehicle measure if the vehicle’s inside tires lift off the pavement during the maneuver ("inside" meaning the left wheels if turning left, and the right wheels if turning right).  The vehicle is considered to have tipped up in the maneuver if both inside tires lift at least two inches off the pavement simultaneously.  The tip-up/no tip-up results are then used with the SSF measurement as inputs in a statistical model that estimates the vehicle’s overall risk of rollover in a single-vehicle crash.[13]

Thus, for model years prior to 2004, rollover resistance ratings were based solely on the SSF.  Starting with 2004 model year vehicles, rollover ratings combine both SSF and the tip-up or no tip-up results of the dynamic maneuvering rollover test.  The results of this new process of generating rollover ratings, and its effect on future SSF trends as well as the design of future vehicle, are outside the scope of this report.  They can be evaluated in the future as data become available.

NHTSA’s web site and Buying A Safer Car brochure provide rollover resistance information as well as ratings for specific make and models of vehicles by model year.  Rollover ratings for individual vehicles can be viewed by visiting www.safercar.gov.  Passenger vehicles of all types (cars and LTVs) are included.  For simplicity, ratings are converted to a graphical star system, similar to NCAP’s impact testing program.  A vehicle’s rollover resistance rating is an estimate of its risk of rolling over in a single-vehicle crash, not a prediction of the likelihood of a crash. As the chart below indicates, the lowest-rated vehicles (1 star) are at least four times more likely to roll over than the highest-rated vehicles (5 stars) when involved in a single-vehicle crash.

Source: http://www.safercar.gov/Rollover/pages/RatSysInterpret.htm

Because rollover risk is a function of vehicle design, the different vehicle types tend to have differing patterns of risk.  The injury and fatality experience also bear this out, as previously noted.  The chart below helps consumers make comparisons across vehicle classes. The information is based on data from all vehicles tested under NHTSA’s 2001-2003 rollover resistance ratings system (SSF only).  For example, all passenger cars that were tested have 4- or 5-star ratings, and an average chance of rollover of approximately 12% if involved in a single-vehicle crash. The range for passenger cars is approximately 6% to 19%.

Source:   http://www.safercar.gov/Rollover/pages/RatSysVCompare.htm

The purpose of this report was to track the trend in SSF over several years for numerous vehicles, to determine whether more recent models have higher SSF values and are therefore less prone to rollover.  The rollover ratings are produced for consumer information only; no regulatory requirements specifically related to rollover mitigation are placed on vehicle manufacturers.  If consumers use the information to purchase vehicles with higher rollover reliability ratings, then manufacturers would presumably design and produce vehicles to meet that public demand.  This report attempts to see if that is what has indeed occurred in the marketplace.

Data Acquisition

In order to examine the trends in the static stability factor over time, it was necessary to locate SSF values for as many different vehicles, over as many model years, as possible.  Several published sources of data ^{[14],[15],[16],[17]} provided such data over a number of years.  More recent information was obtained from NHTSA’s NCAP data.

An SSF value for a given model year was considered to be the same for all model years of a particular vehicle during which it had not undergone a major redesign.  For example, an SSF measurement of 1.334 was obtained for a 1987 Toyota Camry.  The Camry was first introduced in model year 1983.  Although numerous changes were made over the years (including noticeably increased power and a sleeker look in 1987), it wasn’t until 1992 that there was what would be characterized as a major redesign.  (An example of a major redesign would include a change in wheelbase – for the Camry this remained at 102.4 inches until model year 1992, when it increased to 103.1.)  Thus, the value of 1.334 was used for all model year 1983 through 1991 Toyota Camrys.  Another SSF measurement, 1.460, was obtained for the model year grouping 1992 through 1996 (after which there was another major redesign), and this value was used for this set of model years.

If more than one SSF value was obtained for a model year grouping, the midpoint model year between two consecutive values would be determined, with the earlier value being used for all model years earlier than that point, and the later value for the more recent years.  If two distinct SSF measurements were obtained for the same model year, these were averaged.  In some cases, however, while the SSF value was known for a model year grouping, it could not be determined which model year vehicle had been measured.  In this case, the value itself was used for the appropriate span of model years.  However, if a second value was obtained for a vehicle in that model year grouping, the values were averaged, since a specific model year was not known for one of the values.  This was the procedure used to assign SSF values to each model year for a particular vehicle.

Along with covering a span of model years, vehicles that were clones (“sisters”, “corporate cousins”) of one another were also considered to have the same SSF value in cases where the vehicles were virtually identical.  These sister vehicles are predominantly comparable, being built on the same platform.  Measurements of vehicle height and width were used to determine whether specific models were alike enough to be considered as having the same SSF value, since these would directly affect the SSF.  For example, one group of corporate cousins consists of the Dodge Daytona, model years 1984 through 1993; Plymouth Sundance, 1987-1994; Dodge Shadow, 1987-1994; and the Chrysler Laser, 1984-1986.  An SSF measurement was available for the 1987 Plymouth Sundance.  While the four vehicles all have a wheelbase of 97 inches and are built on the same platform, only the Plymouth Sundance and the Dodge Shadow had identical height and width measurements (52.7 and 67.3 inches, respectively).  The Plymouth Laser, however, had a height of only 50.3 inches, and a width of 69.3 inches.  The Dodge Daytona’s height was 51.8 inches, with a width of 69.3 inches.  Thus, the SSF for the Sundance was also used for the Shadow, over the full range of model years, but not for either the Laser or Daytona.

In order to determine the trend in SSF in the motor vehicle fleet over time, it was necessary to examine the data weighted by vehicle sales.  Thus, vehicles that represented a larger portion of the fleet would correspondingly contribute more to the SSF value for the model year.  Registration data from the National Vehicle Population Profile[18] (NVPP) was primarily used to approximate the number of vehicle sales.  The number of vehicles of a given model year registered in the following calendar year (e.g., the number of model year 2001 Pontiac Bonnevilles registered in calendar year 2002) is a good estimate of total sales, because by the following year nearly all the vehicles of the model year have been sold, and very few have already been retired.  This registration was supplemented with registration and sales information from several years of the Ward’s Automotive Yearbook[19].

Exhibit 1 presents, for data used in the analysis, the percentage of each type of vehicle for which both SSF values and sales (approximated from the next year’s registration) were obtained.  In all cases SSF values were available for earlier model years, but did not provide a large enough sample to be considered representative.  At least 35% of the fleet was required in order for the data of that model year and vehicle type to be used.  The one exception is full-sized vans, typically more of a commercial vehicle than the other types, which are more consumer oriented.    This could account for their lack of available historical SSF values.

The SSF measurements for full-size vans that were available were for those vehicles more likely to be owned and used by individuals rather than commercially.  For example, Ford E150 vans were used in the study, but there were no SSF measurements available for E250 or E350 vans.  In order to include the full range of passenger vehicles, it was decided to retain the full size vans, keeping them as a separate group and noting the small sample size.  Minivans, however, are well represented, as are all other passenger vehicle types.  Knowing that such a large proportion of vehicles were included in the analysis lends confidence that the findings are indeed representative of the fleet for each model year.

Exhibit 1:  Percent of Each Vehicle Type with SSF and Sales Data, by Model Year

Changes in SSF Over Time

The objective of this report was to monitor the trend in static stability over time.  The average SSF value was determined by vehicle type for each model year, weighted by vehicle sales.  Thus, a passenger car with sales of 100,000 units would influence the average SSF twice as much as a vehicle selling 50,000.

Exhibit 2 presents the average SSF, by vehicle type, over the model years for which data were available.  Several interesting trends are apparent.  Passenger cars have had very stable SSF values over the years.  Note the slight lowering of SSF scores throughout the 1980s, possibly due to the concurrent trend of vehicle downsizing.  However, passenger car scores have not only recovered but actually improved over the years.  SUVs have consistently and substantially improved their scores over time.  More specifically, there was only slight variation in SSF values for SUV from as far back as 1978 through 1998.  In 1999, the average SSF for SUVs reached higher than it ever had, and has been rising since then, particularly in the last three years.  Pickup trucks have maintained consistent SSF values over the full span of years of available data.  Minivans have shown considerable improvement since they were first introduced in the mid-1980s.  Full-size vans, on the other hand, have had little change, but did show small, consistent improvement.  

Exhibit 2:  Average SSF by Vehicle Type, by Model Year                                            (Weighted by Vehicle Sales Data)

Exhibit 3 presents the number of different make/models that contributed to the average SSF values presented above.  This was not based on registered vehicles or sales data, but the number of unique vehicle make/models that were incorporated into the data.  Corporate cousin vehicles were counted as individual vehicles in the table, since they are individual models that a consumer would consider purchasing.  In addition, where data were available for both two-wheel and four-wheel drives, these were treated as separate vehicles.  Some of the numbers of vehicles here are quite low, but still cover a large portion of the total vehicles sales (see Exhibit 1).  These data are presented to provide information on the variety of vehicles used in the analysis.

Exhibit 3:  Number of Vehicles used to Determine Average MY SSF

While looking at the SSF of the full fleet of vehicles each model year is quite useful, examining scores for vehicles as they are introduced gives a picture of the changes that occur each year as new vehicles are established.  That is, what is the average SSF value for each vehicle type, looking only at the year a specific make/model is introduced or redesigned?  This gives a more immediate picture of how vehicles are changing, but does not take into account that the design may stay the same in subsequent model years.  For example, the 1983 through 1991 Camry (see above) had an SSF of 1.334, and the redesigned Camry in 1992 had an SSF of 1.460.  The value of 1.334 would contribute only to the passenger car average of 1.31 for the model year 1983 in Exhibit 4, while the value of 1.460 would be a factor only in the 1992 average of 1.44.  Table entries of N/A signify there were no available data on vehicles introduced or redesigned that model year.  These data are presented in Exhibit 4.

Exhibit 4:  Average SSF by Vehicle Type, by Model Year Introduced

(Weighted by Sales Data)

Vehicles manufactured, unchanged, over a number of years would contribute to each model year it was available for Exhibit 2, but only once for Exhibit 4.  Thus, the data in Exhibit 4 are based on much smaller samples than those in Exhibit 2, and it is not surprising that the SSF values are more variable in Exhibit 4.

Passenger cars show the drop in SSF values from the early to late 1980s that was also seen in Exhibit 2, along with the following increase.  The SSF measurements for passenger cars have remained high and comparatively consistent since then, although the trend after 1997 indicates newly designed models look to have slightly lower SSF values than models introduced in the early 1990s.  SUV introductions have relatively consistent SSF values for most of the model years with available data, but do show an increase from about 1998 onward.

Newly introduced pickup truck SSFs range from 1.07 to 1.25, with no discernable pattern.  It is clear, though, from comparing these data to those in Exhibit 2, that those newly introduced pickups with low SSF scores are offset by other available models and do not substantially affect the consistent model year average.

Minivans have shown substantial, consistent improvement in SSF values over the most recent decade.  The average SSF of those introduced in 1995 through 2002 was about 0.10 higher than the preceding decade (an increase from 1.13 to 1.23).  Little data are available for full van introductions, certainly not enough from which to draw inference.

The recently introduced “crossover” utility vehicles (CUV) are the most consistently stable SUVs that have been manufactured.  Real improvement is seen in this group of vehicles.  Exhibit 5 presents SSF information on SUVs, separately for standard and crossover models.  The original data for all SUV are repeated from Exhibit 2 for reference purposes.  1997 was the first model year with available data for any crossover SUV, so from that point on SSFs are presented separately for standard and crossover vehicles as well as for all SUVs combined.

Exhibit 5:  Average SSF by SUV Type, by Model Year

Note that SSF values for crossover SUVs were clearly higher than standard SUVs since their introduction.  In addition, standard SUVs showed no change in SSF for a number of years, but by model year 2002 were showing a noticeable increase.  Even the SSF values for crossover vehicles, which started relatively high, have increased over time.  Thus, the SSF values for SUVs have increased, due to both the introduction of crossover vehicles as well as increases in SSF for traditional SUVs.  The introduction of crossover vehicles, and their subsequent increase in percentage of fleet vehicles, has been an important factor in the increase in SSF values of SUVs.  Exhibit 6 shows the percentage of the SUV fleet made up of crossover vehicles.  In the first model year crossover vehicles were introduced, 1997, they made up less than six percent of all SUVs.  Six years later, more than one out of every three SUVs sold is a crossover vehicle.

Exhibit 6:  Percentage of new SUVs that are Crossover Vehicles

Exhibit 7 presents the percentage of each type of vehicle (standard and crossover, as well as all SUVs) for which data were available.  For crossover vehicles, the number of models with available SSF information is also listed.  Note that for model year 1997, only one model was used, but accounted for 58 percent of crossover vehicles.  Over time, more models became available, and sales of such vehicles increased as well, as shown in Exhibit 6.

Exhibit 7:  Percent of Each SUV Type with SSF and Sales Data, by Model Year

Number of Crossover models shown in parentheses

Overall, the news is good.  Static stability factor has tended to increase for all vehicles, particularly SUVs, which tended to have the worst SSF values in the earlier years.  Exhibit 2 shows that SUVs have recently raised their SSF values well above those of the full-size vans.  The popularity of SUVs with consumers makes this a most welcome improvement.

SSF Increases in Specific Vehicles

Greater vehicle stability, as evidenced by increasing SSF values across vehicle types, occurs both as new make/models enter the market and as previous models are redesigned to appeal to a larger number of consumers.  In the first category, perhaps the most noteworthy examples are the crossover vehicles that have been introduced in recent years.  These are vehicles that have characteristics of two vehicle types, typically combining aspects of a passenger car with a sport utility vehicle.  Examples would be  the “corporate cousins” Ford Escape and Mazda Tribute, both introduced in model year 2001.  These vehicles (classified as SUVs in this report) had SSF values of 1.23.  The introduction of this new type of vehicle has been beneficial in increasing the SSF of SUVs.

A redesigned vehicle can be compared to its earlier model when SSF values are available for both versions, to note improvements that have occurred.  Exhibit 8 presents some examples of passenger cars that increased their SSF value by 0.09 or more as part of a major redesign, while Exhibit 9 contains the same information on LTVs.  Note that this report concerns vehicles

Exhibit 8:   Passenger Cars Redesigned with Increased SSF

Exhibit 9:   LTVs Redesigned with Increased SSF

through model year 2003 only; thus that is the last possible model year listed in the Exhibits.  Note also, in some cases what is listed as a redesigned vehicle is actually a successor vehicle, replacing a vehicle the manufacturer stopped producing.  For example, the Ford Explorer is the successor to the Ford Bronco II.  (There may exist other examples, but these are the ones for which NHTSA has before-and-after SSF test results.)

The previous listed examples are all of consecutive (and for some successor vehicles, concurrent) model year redesigns resulting in increased SSF values.  Other examples are available in which an earlier vehicle has a substantially lower SSF than its redesigned counterpart, but SSF data are unavailable for some interim period.  For example, the SSF for the 1983-1988 Ford Thunderbird is available.  No measurement is available for the redesigned version produced from 1989 through 1997.   The Thunderbird was not in production from 1998 through 2001, but in 2002 reappeared on a new platform, having undergone a major overhaul.  Exhibits 10 and 11 present, for passenger cars and LTVs, respectively, examples of these interrupted or “long term” increases in SSF.

Exhibit 10:   Passenger Cars Redesigned with Long-term Increased SSF

Exhibit 11:   LTVs Redesigned with Long-term Increased SSF

Of course, not every new vehicle has a higher SSF than those that preceded it.  If that were the case, the average SSF would have increased even more sharply than seen in Figure 2.  The vehicles noted in Exhibits 8 through 11 are presented as some outstanding examples of passenger cars and LTVs that have been redesigned and/or succeeded by vehicles offering substantially improved stability.

As seen in this report, vehicles today are considerably improved with respect to stability as compared to those of the past, particularly in the case of sport utility vehicles.  By providing information on SSF, NHTSA has enabled the consumer to make a better-informed purchase.  Given the lead-time necessary to introduce a new vehicle or redesign an existing vehicle extensively enough to alter its SSF, it is unlikely that market incentives begun in model year 2001 (when NCAP consumer rollover ratings were first available) could have influenced SSF values by model year 2003.  However, it seems that by model year 2003, both manufacturers and consumers were in agreement of the need for greater stability in passenger vehicles.  Examining trends in SSF over an even longer period of time would enable a more definitive statement on the impact that published NCAP information has had.  However, the evidence suggests that manufacturers are responding to the marketplace and incorporating desirable changes into the new vehicle fleet.