Losing Weight to Save Lives:

F O R U M O N P H Y S I C S & S O C I E T Y
of The American Physical Society
January 2002



Previous Newsletters

this issue

Contact the Editors

Losing Weight to Save Lives: A Review of the Role of Automobile Weight and Size in Traffic Fatalities

Marc Ross and Tom Wenzel

This is a short version of a 3/13/01 report to the National Research Council’s Committee on Effectiveness and Impact of Corporate Average Fuel Economy (CAFE) Standards,.*

Critics of higher fuel economy standards argue that improving fuel economy will require reducing vehicle weight, and that that would result in more deaths. In this report we study the safety implications of improving fuel economy in a more sophisticated way than across-the-board mass/size reduction.

Fatality statistics, 1999. The Fatality Analysis Reporting System (FARS) produced annually by the National Highway Traffic Safety Administration (NHTSA) includes a record on every fatal highway crash, with about 340 variables for each.

Table 1 shows the distribution of deaths by type of crash. If one excludes pedestrian and bicyclist deaths, essentially half of all fatalities are the result of collisions between two vehicles, while the other half are the result of one-vehicle crashes. The latter includes crashes due to the driver’s loss of control rather than to a collision with an object.

Table 1. Fatalities by Type of Crash, 1999 FARS, as Defined by "First Harmful Event"


one-vehicle w/ objectb









a) Of these, there were 6193 deaths in car-to-light truck crashes and 1289 in motorcycle-to-vehicle crashes.

b) stationary objects like trees, guard rails, and utility poles, but not temporarily stopped vehicles

c) 80% of these are primary rollovers.

The advantage for occupants of heavy vehicles is evident in Table 2. For instance, buses and heavier trucks are 9% of vehicles in crashes, but their occupants account for only 2% of fatalities.

Table 2. Vehicle Types Involved in Fatal Crashes, 1999 (pedestrian & bicycle deaths excluded)



light trucks

other trucks, buses


Vehicles, total involved





Fatalities to vehicle occupants





The relatively high rate of light-truck rollover fatalities can be seen in Table 3. Kahane’s (1997) regression analysis also found that weight reduction in cars would increase the rollover risk, because, for 1980s model years, light weight was correlated with fatal rollovers. But this tendency largely disappeared in the 1990s as manufacturers improved car designs. The correlation of rollovers with weight is not inherent, but depends on features such as height of center of mass, track width, and stiffness of suspension. With rollover standards, changes will also be made to reduce rollovers for light trucks.

Table 3. Number of Vehicles in Fatal One-Vehicle Crashes, by "First Harmful Event", 1999


Light trucks


One-vehicle collision with objects








Seat belts are the most important safety technology. Their importance is evident from comparing belt use by drivers killed in crashes with that observed in the careful National Occupant Protection Use Survey, Table 4. Seat belt use among drivers killed is roughly half as great as among all drivers! Part of the reason is that careless drivers, who are more likely to be involved in a crash, are less likely to use seat belts.

Table 4. Seat Belt Use by Drivers in Percent

Of Those Killed in Fatal Crashes (FARS 1999)

In a Survey of All Drivers (2000)

Passenger Cars



Vans and SUVs



Pickup Trucks



Historical trends. The total number of fatal crashes and fatalities declined 18% from 1979 to 1999, even though the number of vehicles in use increased 55% and estimated vehicle miles traveled increased 75%. There was a dramatic decline of fatalities in car-to-car collisions, accompanied by a sharp rise in the fatalities involving light trucks. While car-to-car crashes became much safer, the number of light trucks and dangerous car-to-light truck crashes increased.

Analysis of fatalities in one- or two-year-old cars which crash with cars of any age reveals patterns for late-1990s models. Dividing by the number of new cars, one obtains the deaths in car-to-car crashes per million new cars. We find the extraordinary result that deaths in new cars from car-to-car head-on collisions were only a third as great in 1998 as one decade earlier, and only 20% as great as in 1980. The improvement shown in Table 5 is partly due to standardized testing programs which have emphasized crash types in the order shown. These developments suggest a point of departure: Driving cars in an environment of cars has become much safer, and there are great opportunities for further progress. Before building on this concept, consider some background.

Table 5 Decline in Driver Fatalities per Car, for Cars One and Two Years Old, 1988 to 1998

Kind of Crash




car-to-car head-on

new cars


car-to-car side impact

new cars


car collision w/ stationary objects

new cars


Exposure. Qualitatively, the risk of a fatality is the product of the risk of a crash, and the risk of fatality given that crash:

R(fatality) = R(crash)•R(fatality|given the crash) (1)

Often analysts choose the total number of vehicles as the denominator or the measure of "exposure" to a fatal crash. The exposure is then the number of registered vehicles. Consider, however, young male drivers. Teenage male drivers are so much more likely to crash that they have a fatality risk about 4 times that of 35-to-50 year old males, and 7 times that of 35-to-50 year old females, even though they are less fragile (Kahane 1997, p 6). Consider driving on rural roads, excluding expressways. In a survey of 1999 fatalities at the county level, we find a striking pattern: several times more deaths per resident in counties with low density (Scientific American, Aug. 1987, for 1980). The uncertainties inherent in exposure bedevil interpretation of crash data.

Standardized crash tests. The crashworthiness of new vehicle designs is regulated through Federal Motor Vehicle Safety Standards, and publicized through a rating program, the New Car Assessment Program (NCAP). Both are managed by NHTSA and based on standardized crash tests. The long-running test is for frontal collisions. New vehicles are equipped with instrumented dummies and crashed head-on into a fixed barrier, at 30 mph for the regulation, and 35 mph for NCAP.

A striking but not so surprising consequence of the measurements reported under NCAP are the great reductions which have occurred over time in frontal-collision ratings for many vehicle models. The capability of engineers to improve products so they meet challenging requirements in standardized tests is well established. These improvements are partly responsible for the declining fatality rates in FARS.

Elementary physics. In moderately severe head-on crashes, one may be able to neglect the occupant’s contact with hard objects, and analyze the deceleration of the vehicle. The collision of two vehicles tends to leave them attached. In the simplest case, conservation of momentum shows the lighter vehicle, L, experiences a larger change in velocity than the heavier vehicle, H: DvLDvH = mH/mL (2)

The average deceleration to which the vehicle is subject, <a>, is approximated

< a > = v c Dv/[2(s L + s H )] (3)

where sL = the crush distance of vehicle L, and v c is the closing speed. Eq(2) shows that a light vehicle is less safe in collision with a heavy vehicle. Eq(3) shows that a vehicle with a small crush space is less safe. However, the relative roles of mass and space are not understood.

Peak deceleration correlates with severity of injury. One way to mitigate peak deceleration is to lengthen crash duration by increasing crush space. (A serious collision lasts roughly one-tenth of a second.) Another is vehicle design that spreads the deceleration more evenly during the crash. However, all fatalities cannot be eliminated; and one cannot usually protect adequately at high speeds.

Car-to-car frontal collisions. Consider the ratio of the fatality risks of drivers in two-vehicle collisions (Joksch 1998, section 5). This ratio doesn’t involve the likelihood of collision. To separate the effects of gross vehicle structure from vehicle mass/size, we consider car-to-car head-on collisions. As suggested by Eq(2), the risk of death depends empirically on the ratio of the car masses. Joksch finds for early 1990’s crashes: RL/RH ~ (mH/mL)n with n ~ 3 (4)

This means (for an average case) that if car H is 1.6 times as heavy as car L, then 4 times as many are killed in car L as those in car H. (For current four-door sedans, Cadillac Seville and Honda Civic, the mass ratio is 1.58.) Improving safety technology has affected these mass-risk relations. For example, average results for model years 1995-99 imply a fatality risk ratio of 2.3, rather than 4, for a mass ratio of 1.6. Of course, this is still a large ratio. Although presented in terms of mass, these are dependences on mass and correlated dimensions.

The risk to passengers in both vehicles. While it is more risky to be in a light/small vehicle than in a heavy/large vehicle if the two collide, let us ask: What is the risk to society considering the occupants of both vehicles? And how does that risk change as masses/sizes are changed?

In the most convincing analysis, Joksch shows that the dependence of the ratio, RL/RH, just discussed, establishes rather generally that in collisions between two cars of the same mass there are many fewer fatalities than for collisions between cars of different mass, but the same average mass. That is:

RL + RH declines as mH/mL ® 1 (5)

with mH + mL unchanged

This analysis is straightforward and convincing because it is based on fatality ratios.

Cars vs light trucks. Fatalities in car-to-light truck head-on collisions are about 5 times higher in the car than in the light truck (Joksch 1998, Gabler & Hollowell 1998). In collisions where an SUV strikes the left side of a car, there are 30 driver fatalities in the struck car for each fatality in the striking SUV! Today’s light trucks are incompatible with cars.

Table 6 enables comparison of the fatality risk for both drivers in car-to-car and car-to-light truck collisions (Joksch 2000, pp 9-10). This approach shows, for example, that the ratio of fatalities in SUV-to-car collisions to those in car-car collisions is 2.32/1.28 = 1.8. Based on the 1013 driver fatalities in SUV-to-car collisions (Table 7), the ratio 1.8 leads to an estimate that about 450 excess driver’s lives were lost in 1999 due to the use of SUVs as substitutes for cars. On this basis, an annual excess of about 2200 deaths may be associated with the use of light trucks as car-substitutes.

Table 6 Drivers Killed per 1000 Drivers Involved: Car-to-Light Duty Vehicle Collisions,1991-1997. The "denominator" is 1000 times the number of police-reported crashes.

Death in:

Other vehicle type










other vehicle





both vehicles





Table 7 Driver Fatalities in Car-to-Car and Car-to-Light Truck Collisions, 1999 FARS

Death in:

Other vehicle type










other vehicle





both vehicles





Ford recently announced design changes to improve the compatibility of SUVs and pickups with cars, starting in MY2004. This is a promising, if modest, beginning. Presumably others are taking similar steps.

A Safety-Fuel Economy Scenario. In order to address both fuel economy and safety, we propose making heavier vehicles lighter but not smaller, and making lighter vehicles larger but not lighter. These changes would be enabled by mass-reduction technologies. Two or three major kinds of changes could be made to reduce mass, independent of size:

1) The basic structural design of those light-truck car-substitutes which are now "body on frame" could be unibody, like today’s cars, or a skin-on-frame design called space frame. These structures would also improve compatibility between light trucks and cars.

2) Lightweight materials would be emphasized, such as high and ultrahigh strength steels, aluminum and engineering plastics.

3) High-efficiency propulsion systems would be lighter. The technologies could include: a) smaller displacement engines with high ratio of power-to-displacement, b) automatic transmissions without torque converter (with motor-shifted standard transmissions using a sophisticated clutch and management to assure smooth acceleration, or with continuously variable transmission), and c) on-shaft starter-generators, with the imminent 42-volt electrical system, enabling idle-off and other modest hybrid-drive capabilities without a significant battery mass penalty. The shift busyness and slight shift delays that characterize driving with a powerful but small engine could interfere with marketing. But minimizing this problem should be viewed as an engineering challenge. The concept is to achieve smooth control of acceleration through intelligence rather than friction.

To illustrate, we assume that the mass range (curb weights) of most light duty vehicles would be reduced to between 2400 and 3300 lbs (the present distribution is 2400 to 5000 lbs). Mass reductions of up to 33% are projected. About half of light-duty vehicles would be reduced in mass, with an average 1000 lb reduction. To achieve such major mass reductions while maintaining the size and performance characteristics that attract customers would be a major engineering challenge. In addition, safety technologies would continue to be developed and applied.

We estimate this scenario would save about 2200 lives a year in two-vehicle crashes. This estimate is conservative, based on the excess of deaths in light truck-to-car crashes, for only those light trucks used as car substitutes, and using 1999 fatality rates. There would also be many lives saved in car-to-car crashes, but some extra lives lost in heavy truck-to-light duty vehicle crashes. We also estimate an increase of less than 400 deaths per year in one-vehicle crashes with stationary objects.

We project improvement in combined city-highway fuel economy of 55% (DeCicco et al. 2001). Fuel economy improvement in gas guzzlers is especially important for fuel savings and CO2 reduction.

There would be increases in manufacturing cost-per-unit in this scenario if the lightweight materials aluminum and engineering plastics are substantially involved. However, ultralight steel techniques are not costly, and can reduce the mass of steel bodies and associated parts 15% to 20%. While there would be substantial development and re-tooling costs, distinct from the per-unit manufacturing costs, in the larger picture of automotive model changes and safety they appear reasonable (DeCicco et al. 2001).

Marc Ross

Physics Dept, University of Michigan, Ann Arbor, MI 48104-1120

734 764-4459

Tom Wenzel

Lawrence Berkeley National Laboratory, Berkeley, CA 94720

510 486-5753, TPWenzel@lbl.gov



DeCicco, John, Feng An & Marc Ross, Technical Options for Improving the Fuel Economy of US Cars and Light Trucks by 2010-2015, American Council for an Energy-Efficient Economy, Washington DC, 2001.

Gabler, HC, & WT Hollowell, The Aggressivity of Light Trucks and Vans in Traffic Crashes, Society of Automotive Engineers, 980908, 1998.

Joksch, Hans, Vehicle Design versus Aggressivity, NHTSA DOT HS 809 194, April 2000.

Joksch, Hans, Vehicle Aggressivity: Fleet Characterization Using Traffic Collision Data, NHTSA, DOT HS 808 679, 1998.

Kahane, Charles J, Relationships Between Vehicle Size and Fatality Risk in Model Year 1985-93 Passenger Cars and Light Trucks, NHTSA, DOT HS 808 570, 263 pages, 1997.




Previous Newsletters

this issue

Contact the Editors