IV.  TARGET POPULATION

Safety Problems associated with Low Tire Pressure

Under-inflation affects many different types of crashes.  In commenting to the docket, the International Tire & Rubber Association (ITRA) (Docket No. 8572-123) stated that when developing ITRA training programs they look closely at tire performance and have the opportunity to analyze a significant number of tires that failed in service.  They find that the single most common cause of tire failure is under-inflation. 

The types of crashes that under-inflation influences are:

1. skidding and/or a loss of control of the vehicle in a curve, like an off-ramp maneuver coming off of a highway at high speed, or simply taking a curve at high speed
2. skidding and/or loss of control of the vehicle in a lane change maneuver,
3. hydroplaning on a wet surface, which can affect both stopping distance and skidding and/or loss of  control. 
4. an increase in stopping distance,
5. crashes caused by flat tires and blowouts
6. overloading the vehicle

We can identify target populations for skidding and loss of control crashes, flat tires and blowouts, and stopping distance (which involves any vehicle that brakes during a crash sequence).  We cannot identify from our crash files, or other reports, the incidence of hydroplaning specifically (we do however identify wet surfaces and loss of control in our "skidding and loss of control" analysis of crashes), or the impacts of overloading a vehicle (this may be captured somewhat in tire blowouts). 

Skidding and loss of control

The 1977 Indiana Tri-level study associated low tire pressure with loss of control, on both wet and dry pavements.  They never identified it as a "definite" (95 percent certain that the crash would not have occurred without this cause) cause of any crash, but did identify it as a "probable" cause (80 percent confidence level  - highly likely that the crash would not have occurred) of the crash in 1.4 percent of the 420 in-depth crash investigations.^[1]

Probable cause was broken up into two levels: a causal factor and a severity-increasing factor.   A causal factor was defined as "had the factor not been present in the accident sequence, the accident would not have occurred."  A severity-increasing factor was not sufficient to result in the occurrence of the accident, but resulted in an increase in speed of the initial impact.  Under-inflated tires were a causal factor in 1.2 percent of the probable causes and a severity-increasing factor in 0.2 percent of the probable causes.   

Note that more than one "probable cause" could be assigned to a crash.  In fact, there were a total of 138.8 percent causes listed as probable cause (92.4 percent human factors, 33.8 percent environmental factors, and 12.6 percent vehicular factors).  Thus, under-inflation’s part of the total is 1.0 percent (1.4/138.8).  If we focus on just the probable cause cases, under-inflation represents 0.86 percent of crashes (1.2/1.4*1.0).    

There are several important factors to know about the Indiana Tri-Level study and their implications for this analysis.  This information was verified with the authors of the study and NHTSA contract technical managers on the study.

1. None of the cases in which under-inflation was cited as a probable cause dealt with stopping distance.  They were all cases of loss of control in a curve or in a crash avoidance maneuver.

2. High speed was not a factor in these cases.  In order to be considered for an under-inflation case, the vehicle had to be going within a reasonable speed to make the turn, for example.

3. In order for under-inflation to be cited in the study, there had to be a significant amount of under-inflation, 10 to 15 psi low or more compared to placard levels.  Thus, the estimates would apply to all three Alternatives fairly equally.

4. There were particular vehicles that were known to lose traction when their tires were under-inflated in particular patterns, sometimes the rear tires, or sometimes a disparity in inflation.  The authors particularly noted the Chevrolet Corvair and the early-60’s Volkswagen Beetle.  Problem vehicles like this were not a big part of the sample but raised the rate somewhat and do not appear to be a problem today.  We assume this factor could reduce the probable cause estimate by 10 percent to 0.77 (0.86*.9).

At the time of the study, radial tires were on 12% of passenger vehicles, and now they are on more than 90% of passenger vehicles, including all tires on new automobiles.  The question is whether the 1977 results are applicable in today’s tire environment.  The agency at this time is unable to quantify how the cornering force capability of different tire constructions (bias ply, bias belted and radial) at different tire inflation pressures affects the frequency of loss of control crashes. Radial tires provide better tread contact with the pavement since their sidewalls are more flexible in the lateral direction than bias ply tires. Accordingly, radial tires can generate about twice the lateral force as bias ply tires.  However, drivers get feedback from their tires and drive vehicles with different types of tires in different ways around corners.  Bias and bias belted tires provide more feedback to the driver by feel and noise that the vehicle might not negotiate a curve, and the driver can sometimes slow down and correct the situation before going off the road.  While radial tires generate more lateral forces, they do not provide progressive feedback to the driver and tend to lose traction without as much warning.  In essence, drivers have learned how to go around entrance and exit ramps, and other curves, on highways at a higher rate of speed with radial tires.  However, if the road is wet and their tire pressure is low, then they might have problems taking that curve at the same speed.  Thus, we can’t determine how to correct the Indiana Tri-Level study to account for the difference in types of tires.  It may well be, and for this assessment we assume, that the same percentage of under-inflation influenced crashes occur with radial tires as with bias and bias-belted tires.

To get an estimate of the target population of the low tire pressure cases in which skidding and loss of control could be a factor, we took data from "Traffic Safety Facts, 1999" which shows there were about 47,848 passenger vehicles (passenger cars and light trucks) involved in fatal crashes, about 3.6 million passenger vehicles involved in injury crashes and about 6.9 million passenger vehicles involved in property damage only crashes.   These crashes resulted in 32,061 passenger vehicle occupants being killed and almost 3 million passenger vehicle occupants being injured.     

Taking 0.77 percent of these cases, loss of control and skidding due to low tire pressure would account for an estimated 247 occupants killed, 23,100 occupants injured, and 53,130 property damage only crashes.

As a second check on these estimates, the 1999 NASS-GES was examined to identify particular crash scenarios in which loss of control occurred.  The following scenarios that could be identified were examined totaling over 413,000 vehicles (3.9 percent of the vehicles in all crashes).  Certainly there are other scenarios that couldn’t be identified, but this check was made to assure ourselves that 0.77 percent was not impossibly high. 

Negotiating a curve: Where the vehicle left the roadway, left the travel lane, lost control or skidded (213,759 vehicles)

Changing lanes where the vehicle left the roadway, lost control or skidded (4,890 vehicles), and

Raining cases where the vehicle lost control and skidded (194,709 vehicles).   

Flat tires and blowouts

There is no direct evidence in NHTSA’s current crash files (FARS and NASS) that points to low tire pressure as the cause of a particular crash.  This is because we have no measurements of tire pressure in our data bases (plans are underway to start collecting this data in 2002).  The closest data element is "flat tire or blowout".  Even in these cases, crash investigators cannot tell whether low tire pressure contributed to the tire failure.  Tire failures, especially blowouts, are associated with rollover crashes.  Low tire pressure can also lead to loss of control or a skid initially.  Skids can lead to tripping and then to a rollover.

The agency examined its crash files to gather whatever information is available on tire-related problems causing crashes.  The National Automotive Sampling System - Crashworthiness Data System (NASS-CDS) has trained investigators that collect data on a sample of tow-away crashes around the country.  These data can be weighted up to national estimates.  The NASS-CDS contains on its General Vehicle Form the following information: a critical pre-crash event, vehicle loss of control due to a blowout or flat tire.  This category only includes part of the tire-related problems causing crashes.  It does not include cases where there was improper tire pressure in one or more tires that did not allow the vehicle to handle as well as it should have in an emergency situation.  This coding would only be used when the tire went flat or there was a blowout and caused a loss of control of the vehicle, resulting in a crash.  However, as stated above, low tire pressure may contribute directly to the crashes discussed in the paragraphs below.  In addition, there may be other crashes, not included in the paragraphs below, where low tire pressure played a part.

NASS-CDS data for 1995 through 1998 were examined and average annual estimates are provided below in Table IV-1.  Table IV-1 shows that there are an estimated 23,464 tow-away crashes caused per year by blowouts or flat tires.  Thus, about one half of a percent of all crashes are caused by these tire problems.  When these cases are broken down by passenger car versus light truck, and compared to the total number of crashes for passenger cars and light trucks individually, it is found that blowouts cause more than three times the rate of crashes in light trucks (0.99 percent) than in passenger cars (0.31 percent).  When the data are further divided into rollover versus non-rollover, blowouts cause a much higher proportion of rollover crashes (4.81) than non-rollover crashes (0.28); and again more than three times the rate in light trucks (6.88 percent) than in passenger cars (1.87 percent).

Table IV-1
Estimated Annual Average Number and Rates of
Blowouts or Flat Tires Causing Tow-away Crashes

	Tire Related Cases	Percent Tire Related
Passenger Cars Total	10,170	0.31%
Rollover	1,837 (18%)	1.87%
Non-rollover	8,332 (82%)	0.26%
Light Trucks Total	13,294	0.99%
Rollover	9,577 (72%)	6.88%
Non-rollover	3,717 (28%)	0.31%
Light Vehicles Total	23,464	0.51%
Rollover	11,414 (49%)	4.81%
Non-rollover	12,049 (51%)	0.28%

Table IV-2 shows the estimated number of fatalities and injuries in those cases in which a flat tire/blowout was considered the cause of the crash^[2]. There are an estimated 414 fatalities and 10,275 non-fatal injuries in these crashes.

Table IV-2
Injuries/Fatalities in Crashes Caused by
Flat Tire/Blowout

	Non-fatal AIS 1	Non-fatal AIS 2	Non-fatal AIS 3	Non-fatal AIS 4	Non-fatal AIS 5	Fatalities
Number of Injuries	8,231	1,476	362	155	51	414

The Fatality Analysis Reporting System (FARS) was also examined for evidence of tire problems involved in fatal crashes.  In the FARS system, tire problems are noted after the crash, if they are noted at all, and are only considered as far as the existence of a condition.  In other words, in the FARS file, we don’t know whether the tire problem caused the crash, influenced the severity of the crash, or just occurred during the crash.  For example, (1) some crashes may be caused by a tire blowout, (2) in another crash, the vehicle might have slid sideways and struck a curb, causing a flat tire which may or may not have influenced whether the vehicle rolled over.  Thus, while an indication of a tire problem in the FARS file gives some clue as to the potential magnitude of the tire problem in fatal crashes, it can neither be considered the lowest possible number of cases nor the highest possible number of cases.  In 1995 to 1998 FARS, 1.10 percent of all light vehicles were coded with tire problems.  Light trucks had slightly higher rates of tire problems (1.20 percent) than passenger cars (1.04 percent).  The annual average number of vehicles with tire problems in FARS was 535 (313 in passenger cars and 222 in light trucks).  On average, annually there were 647 fatalities in these crashes (369 in passenger cars and 278 in light trucks).  Thus, these two sets of estimates seem reasonably consistent: 647 fatalities in FARS in crashes in which there was a tire problem and 414 fatalities from CDS, in which the flat tire/blowout was the cause of the crash.    

Geographic and Seasonal Effects

The FARS data were further examined to determine whether heat is a factor in tire problems (see Table IV-3).  Two surrogates for heat were examined: (1) in what part of the country the crash occurred, and (2) in what season the crash occurred.  The highest rates occurred in light trucks in southern states in the summer time, followed by light trucks in northern states in the summer time, and by passenger cars in southern states in the summertime.  It thus appears that tire problems are heat related.

Table IV-3
Geographic and Seasonal Analysis of Tire Problems
 (Percent of Vehicles in) FARS with Tire Problems

	Passenger Cars	Light Trucks	All Light Vehicles
Northern States
Winter	1.01%	0.80%	0.94%
Spring	1.12%	1.01%	1.08%
Summer	0.98%	1.46%	1.15%
Fall	1.04%	0.93%	1.00%
Southern States
Winter	0.87%	0.99%	0.92%
Spring	1.09%	1.27%	1.16%
Summer	1.31%	1.99%	1.59%
Fall	0.89%	1.07%	1.00%

There are also crashes indirectly caused or indirectly involved with tire related problems.  If a vehicle stops on the side of the road due to a flat tire, there is the potential for curious drivers to slow down to see what is going on.  This can create congestion, potentially resulting in a rear-end impact later in the line of vehicles when some driver isn’t paying enough attention to the traffic in front of them.   The agency has not attempted to estimate how often a TPMS would give the driver enough warning of an impending flat tire that they could have the tire repaired before they get stuck having to repair a flat tire in traffic.  However, it should be a very large number.   

An indirectly involved crash relating to tire repairs on the road can occur when someone is in the act of changing a tire on the shoulder of the road.  Sometimes drivers repairing tires are struck (as pedestrians) by other vehicles.  This phenomena is not captured in NHTSA’s data files, but there are three states (Pennsylvania, Washington, and Ohio) which have variables in their state files which allow you to search for and combine codes such as "Flat tire or blowout" with "Playing or working on a vehicle" with "Pedestrians".  An examination of these files for calendar year 1999 for Ohio and Pennsylvania and for 1996 for Washington found the following information shown in Table IV-4.

Table IV-4
State data on tire problems and pedestrians

	Ohio	Washington	Pennsylvania
Pedestrians Injured	3,685	2,068	5,226
Pedestrians Injured While Playing or Working on Vehicle	50 (1.4%)	27 (1.3%)	56 (1.1%)
Pedestrians Injured While Working on Vehicle with Tire Problem	0	2	0
Total Crashes	385,704	140,215	144,169
Crashes with Tire Problems	862 (0.22%)	1,444 (1.03%)	794 (0.55%)

The combined percent of total crashes with tire problems of these three states (3,100/670,088 = 0.46 percent) compares very favorably with the NASS-CDS data presented in Table IV-1 of 0.51 percent. The number of pedestrians coded as being injured while working on a vehicle with tire problems is 2/10,979 = 0.018 percent.  Applying this to the estimated number of pedestrians injured annually across the U.S. (85,000 from NASS-GES), results in an estimated 15 pedestrians injured per year.  It is possible that these numbers could be much higher, if they were coded correctly.  The agency is not going to estimate how many of the pedestrian injuries could be reduced with a TPMS.

[1] Tri-level Study of the Causes of Traffic Accidents: Executive Summary,  Treat, J.R., Tumbas, N.S., McDonald, S.T., Shinar, D., Hume, R.D., Mayer, R.E., Stansifer, R.L., & Castellan, N.J. (1979).  (Contract No. DOT HS 034-3-535). DOT HS 805 099. Washington, DC: U.S. Department of Transportation, NHTSA.  See pages A-51 and D-23 to D-30. 

[2] Since CDS typically underestimates the number of fatalities, a factor was developed based on the number of occupant fatalities in FARS divided by the number of occupant fatalities in CDS for those years of 1.163, which was multiplied by the actual estimate of flat tire/blowout fatalities.