Abstract
This study investigated 478 police accident reports from 9 states to examine and characterize rollover crashes involving ESC-equipped vehicles. The focus was on the sequence of critical events leading to loss of control and rollover, and the interactions between the accident, driver, and environment. Results show that, while ESC is effective in reducing loss of control leading to certain rollover crashes, its effectiveness is diminished in others, particularly when the vehicle departs the roadway or when environmental factors such as slick road conditions or driver factors such as speeding, distraction, fatigue, impairment, or overcorrection are present.
INTRODUCTION
This study addresses the effectiveness of Electronic Stability Control (ESC) systems using field accident data. A recent review of available field data on crash-involved ESC-equipped vehicles indicated that ESC is highly effective for rollovers in which the rollover is the first harmful event (Padmanaban, 2007). However, that study also showed that ESC is not nearly as effective for other types of rollover crashes (Padmanaban, 2007). To further investigate this finding, the present study made a detailed review of police accident reports of single-vehicle rollovers involving vehicles with ESC installed as standard equipment (“ESC vehicles”). In addition to the information coded in state accident data bases, narrative descriptions and scene diagrams were examined to identify factors associated with rollovers involving ESC vehicles.
Past Studies
As defined in Federal Motor Vehicle Safety Standard 126, ESC systems “use automatic computer controlled braking of individual wheels to assist the driver in maintaining control in critical driving situations in which the vehicle is beginning to lose directional stability at the rear wheels (spin out) or directional control at the front wheels (plow out)” (NHTSA, 2007)
By 2011, the standard will require such systems in light vehicles to:
□ Monitor or estimate driver steer input, yaw, side slip and/or side slip rate.
□ Employ a computer algorithm to apply brake torque at individual wheels to produce a corrective yaw moment.
□ And thereby limit oversteer (spin out) and understeer (plow out).
While there are numerous systems, operating under various names, any system referred to in this paper as “ESC” meets the definition provided above. Systems built by diverse manufacturers monitor different vehicle dynamic parameters and may include other interventions as well (such as reduction in engine torque).
Over the past decade, various studies have suggested that ESC systems are effective in reducing crashes; however, most of these studies evaluated very few vehicles (mostly luxury models, since these were the first to be equipped with ESC systems) and used limited crash data and/or vehicle types (mostly cars and SUVs). Mercedes-Benz (2006) examined the effectiveness of their proprietary ESC system in their passenger cars in Germany; Aga and Okada (2003) included only three Toyota passenger car models, involved in traffic accidents in Japan; Lie (2005) included more vehicle model series than either the Toyota or Mercedes studies, but focused on cars only. A recent study of 50,000 accidents in Britain covered more types of crashes and conditions; in that study, ESC effectiveness was estimated to be 7% for all collisions, 25% for fatal collisions, and 36% for rollover events (Thomas and Frampton, 2007).
Several US studies covered a greater variety of vehicles and crashes, but came to differing conclusions. Farmer’s (2004) Insurance Institute for Highway Safety study looked at police-reported crashes (all crash types) in seven states over a two-year period, and fatal crashes in the US over a three-year period, for cars and sport utility vehicles (SUVs) and credited ESC systems with a 41% reduction in all (and a 56% reduction in fatal) single-vehicle crashes, but found reductions for multiple-vehicle crashes to be insignificant. A Pacific Institute for Research and Evaluation study found a 53% reduction in single-vehicle crashes, and a 12% reduction in multiple-vehicle crashes, involving light vehicles (Bahouth, 2005).
A University of Michigan Transportation Research Institute (UMTRI) study examined cars and SUVs separately for crashes “generally associated with loss of control” (Green and Woodrooffe, 2006). The UMTRI study used General Estimates System (GES) files to examine all crashes and Fatality Analysis Reporting System (FARS) data to examine fatal and fatal rollover crashes. The study estimated ESC systems would reduce odds of fatal rollover crashes by 40% for cars and 73% for SUVs. A simulator-based study of 120 subjects in loss-of-control situations with and without ESC systems (Papelis et al., 2004) expressed a more positive view of the system’s usefulness in reducing crashes. That study found ESC systems offered an 88% reduction in loss of control, and its authors concluded that: “In all cases, there was a benefit to having the system” (Papelis et al., 2004).
The National Highway Traffic Safety (NHTSA) ESC study (Dang, 2006) examined police-reported crash data from seven states and found ESC to be effective in reducing single-vehicle “run-off-road” crashes by 46% for passenger cars and 75% for light trucks. The NHTSA study also evaluated FARS data for 1997 to 2004 model year passenger cars and light trucks and found ESC systems to be effective in reducing fatal single-vehicle “run-off-road” crashes by 35% for passenger cars and 72% for light trucks. This study was later updated to include more recent field data and the conclusions were consistent with their earlier study (Dang, 2007).
A study done by JP Research (Padmanaban, 2007) showed that ESC systems were highly effective for fatal single-vehicle rollovers in which the rollover was the first harmful event. For light trucks, ESC was not significantly effective for single-vehicle rollovers where the first harmful events included collision with tree/pole/post/guardrail/culvert or ditch/embankment. These types of rollover crashes were primarily off-road rollover crashes, and NHTSA’s study did not examine the effectiveness of ESC in reducing such rollover crashes.
To better understand factors relating to rollovers involving vehicles with ESC systems, it was deemed appropriate to perform a comprehensive review of field data. Hence this study.
Current Study
As indicated, a comprehensive review was made of police accident reports for 478 single-vehicle rollover crashes involving vehicles equipped with ESC systems. The objectives were to examine and characterize rollover crashes involving ESC-equipped vehicles in greater detail than is possible using only the coded information available in state and FARS databases. This included review of engineering issues that relate to ESC effectiveness. In particular, the sequence of critical events leading to rollovers and the interactions between the accident, driver, and environment were examined for ESC vehicles.
METHODS
Vehicle models with ESC systems as standard equipment were identified using multiple publicly available sources and information from vehicle manufacturers. The study included single-vehicle rollover crashes from the state accident data files, and hard copies of police reports containing accident narratives were obtained. A detailed coding of accident scenarios was performed by an engineering team. The coding effort incorporated 47 variables, including those for critical events (departure from road, loss of directional control, impact with an object, and driver’s input); driver factors (impairment, speeding, inattention, distraction, fatigue, and overcorrection); and environmental factors. Finally, an engineering review was undertaken to provide insights into several accident scenarios in which ESC was not highly effective in preventing rollovers.
Vehicle Classification
Vehicle models with ESC systems were identified using the NHTSA Safercar.gov website, the Insurance Institute for Highway Safety website, MSN Auto, and information from vehicle manufacturers.
Accident Data
For the study, all police accident reports for single-vehicle rollover crashes involving 1997–2006 model year vehicles equipped with ESC as standard equipment were requested from 9 states (Alabama, Florida, Idaho, Kansas, Maryland, Missouri, Nebraska, Washington, and Wyoming). These states include a selection of the environmental and demographic conditions available throughout the country. Police coded fields were examined for consistency across the states, and standardized coding fields were developed for the study.
Of the 507 reports received, 29 accidents were excluded due to the fact that ESC was not standard equipment or the crash was not a SVA rollover. Consequently, the study included 478 single vehicle rollovers.
Approach
Two specific objectives of this study were to use detailed information from police reports to (1) identify factors (including driver, vehicle, and environmental factors) associated with the single-vehicle rollovers in this study and (2) determine critical events leading to those rollovers. Major factors associated with different accident scenarios were examined.
Detailed instructions were developed for variables to be coded from the police reports. These included variables that identify:
□ the sequence of critical events, such as departure from roadway (with or without driver input), and loss of directional control (vehicle is not following the directional path commanded by the driver)
□ driver actions or decisions (such as speeding, alcohol involvement, failure to stay in lane, overcorrection, or fatigue/inattention/distraction) that could have influenced crash occurrence/outcome
□ features of the crash environment (such as weather, roadway features and surface condition, and lighting) that might have influenced crash occurrence/outcome.
Procedures were developed to code variables in a consistent format suitable for input to statistical and engineering analysis. Some variables, such as identifying occurrence of skidding, yawing, or driver overcorrection were coded for their occurrence at any time during the accident. Other variables examined chronologically allowed construction of the critical event sequence leading to each of the rollovers. Examples of critical event sequences include:
□ Vehicle departure from road; steering; reentering road; second departure from road
□ Loss of directional control; departure from road; impact
□ Loss of directional control; steering; braking; skidding; departure from road.
Table 1 (Appendix B) provides a detailed description of all codes used to identify critical events and other factors.
RESULTS
Factors Contributing to ESC Rollovers
Critical Sequences
Review and analysis of the 478 police accident reports resulted in four primary categories of single-vehicle rollovers for ESC vehicles, as shown in Figure 1 (Appendix A):
30% (143) of ESC vehicle rollovers involved departure from roadway without any driver input and prior to loss of directional control.
3% (13) involved impacting an object without driver input prior to either road departure or loss of directional control.
31% (148) experienced loss of directional control prior to impact or road departure.
26% (122) departed the road with possible driver input. The records for an additional 52 crashes (11%) did not have enough information to categorize.
The vehicles in the first and second categories that departed the road or impacted without driver input were at risk prior to the opportunity for ESC engagement.
For the 143 cases involving departure without driver input (first category), the subsequent critical events in the rollover crash sequences were examined, as shown in Figure 2. It was found that:
□ 53% (76) impacted or rolled over
□ 27% (39) reentered road
□ 17% (24) experienced loss of directional control
□ 3% (4) experienced some other event.
For the 39 vehicles that reentered the road, 90% involved steering overcorrection and the vast majority (31) experienced a second departure from road (Figure 3). Twenty eight of these 31 experienced loss of control before the second departure.
The 148 cases in the third category involved a loss of directional control prior to impact or road departure. This means the vehicle, while on the road, was yawing or rotating such that it did not follow the intended path commanded by the driver (based on narrative descriptions and scene diagrams). Subsequent rollover occurred either on or off the road.
Of the 122 cases in the fourth category (departing the road with possible driver input), 47% (57) of the case records included information indicating drivers had attempted to steer or brake prior to road departure. The vehicles in this category demonstrate that, in spite of drivers’ attempts to control their ESC-equipped vehicles, these vehicles may still depart the road and roll over.
Driver Factors
Figure 4 presents driver-related factors associated with ESC vehicle rollovers. Speeding clearly dominates as the most frequent driver factor and was involved in 40% of all the rollovers. Fatigue/inattention/distraction is also prominent and was found present in 29% of the single-vehicle rollover events. Overcorrection and alcohol impairment were also significant factors for rollovers involving ESC vehicles, with overcorrection present in 22% of the rollovers and alcohol impairment present in 21% of the rollovers.
For the two most common critical event categories (departure from roadway without driver input and loss of directional control), driver factors contributing to rollovers were somewhat different. For events in which vehicles departed the road without driver input, fatigue/inattention/distraction dominated the list, with speeding next (Figure 5A). When the first critical event was loss of directional control, speeding was the dominant contributing factor and fatigue/inattention/distraction was next (Figure 5B). These values show fatigue, distraction, and inattention to be (together) a more common factor for those cases in which a vehicle departs the road without driver intervention, and speeding to be a more common factor in loss of control accidents.
Environmental Factors
The role of environment, too, was shown to be important (Figure 6). In 47% of the accidents where the first critical event was loss of directional control, the pavement was slippery (either icy or wet). This is consistent with the fact that maximum braking forces (either applied through the pedal or ESC intervention) are diminished by low coefficient surfaces and, hence, ESC will have less stabilizing capacity. In contrast, 90% of the accidents with departure from road as the first critical event took place on dry roads.
DISCUSSION
Although many of the studies of ESC report high percentages of effectiveness, it is clear that numerous rollover accidents still occur for ESC vehicles. This study considered rollover accidents for ESC vehicles only, thus direct comparisons of ESC and non-ESC experience could not be quantified. However, analyses of those events for which ESC did not prevent rollover were still possible. This study showed that many ESC rollover cases involve situations where ESC has little or no reason or opportunity to effectively intervene, or where its intervening brake and throttle controls are limited by the available friction at the tire/road interface.
In 3% (13) of the cases examined for this study, the vehicles rolled due to impact that occurred prior to either loss of directional control or departure from the road. Another 16% (76) impacted or rolled over after departure from road with no driver input or known loss of directional control that might have engaged the ESC system. Hence in 19% (89) of the studied cases, ESC would not have been called upon to intervene until after impact. Clearly, ESC has no effectiveness if it is not engaged, and unlikely effectiveness after significant impact forces, whether impact occurs on or off the road.
Generally, vehicles that depart the road without driver input are “tracking” (not yawing out of control) as they leave the road. Once off the road, the risk of control loss and rollover is compounded by two factors:
Driver surprise or anxiety apparently stimulates overcorrective steering maneuvers, which can induce particularly high yaw moments. This is amplified when tires differentially engage relatively high-coefficient road surfaces and low-coefficient shoulders and off-road surfaces.
Off-road surfaces are more likely to be irregular, sloping, have areas of soft earth or hazards that can provide momentary and unpredictable yaw forces that would be difficult to avoid or compensate for, and are more likely to have tripping hazards known to trigger rollover (Viano and Parenteau, 2003).
Factors such as these would explain the continued occurrence of loss of control and rollover for those vehicles departing the road without driver input.
The 122 vehicles that departed the road after possible driver braking, steering, or change in throttle control were subject to the same two risk factors described above. This category included 46 cases with known driver input and no loss of directional control prior to departure from road; of these there are likely to be some for which ESC was engaged and loss of directional control was successfully avoided prior to road departure. Still, in the presence of the two off-road risk factors, rollover occurred later in the sequence. Likely to be included in this set would be some cases for which ESC was able to prevent loss of directional control but was unable to maintain the driver’s chosen path on the road. Of course, ESC is unable to choose the most prudent path for safety, regardless of its effectiveness for maintaining the driver’s chosen path.
Of the 478 total cases, 55% (265) involved departure from the road prior to known loss of directional control and despite the presence of ESC. The additional risk of rollover after road departure is not currently considered in NHTSA’s development of vehicle tests that characterize rollover potential (Forkenbrock et al., 2003). Given the variability of off-road conditions, designing such tests would be difficult. Yet, these results show that road-departure conditions are clearly important, and ESC is unable to eliminate off-road excursions and rollovers.
All 148 cases in which directional control was not maintained involved accidents in which ESC engagement was likely. Of the 122 where the vehicle departed the road with possible driver input, 57 involved the vehicle leaving the road with known driver input. It is likely that many of these also involved ESC activation. Consequently, 43% (205) of the cases had still resulted in rollover despite probable ESC activation.
Driver factors of fatigue, inattention, or distraction; speeding; and steering overcorrection, as well as driver impairment from alcohol or drugs, were clearly associated with these rollover accidents.
CONCLUSIONS
While ESC is effective in reducing loss of control in certain situations likely to lead to rollover crashes, its effectiveness is diminished in others, particularly when the vehicle departs the roadway or when environmental factors such as slick road conditions or driver factors such as speeding, fatigue/distraction/inattention, alcohol/drug impairment, or over-correction are present.
Conclusions regarding types of rollover accidents for vehicles with ESC as standard equipment include:
□ For about 30% (143) of rollovers, the first critical event was departure from road without driver input (i.e., before ESC was engaged).
□ Another 31% (148) of rollovers occurred when loss of directional control took place before the vehicle departed the road.
□ About 26% (122) of rollovers occurred when vehicles departed the road after possible driver input.
□ About 55% (265) of rollovers involved departure from the road prior to loss of directional control or rollover.
□ About 43% (205) had likely ESC engagement, but still rolled.
Conclusions regarding driver factors include:
□ For those cases where ESC vehicles reenter road prior to rollover, 90% of drivers imparted steering overcorrection and 80% of the vehicles departed the road again
-
□ Speeding, fatigue/distraction/inattention, overcorrection, and alcohol involvement were predominant driver factors associated with these rollovers.
– For departure from road events: fatigue/distraction/inattention, and overcorrection were predominant.
– For loss of directional control events: speeding was predominant.
□ For all of these rollovers, alcohol involvement was also identified as a significant contributing factor (21%).
Conclusions regarding environmental factors include:
□ Environmental factors such as snowy/icy/wet road conditions were significant factors for loss of directional control events.
FUTURE WORK
The current study examined single-vehicle rollover accidents that occurred with ESC vehicles. A similar review and analysis of vehicles without ESC installation is underway. The purpose of that study is to allow comparisons and contrasts to further explain ESC performance based on differences in frequencies of road departure, loss of directional control, experience on wet and icy roads, and driver demographics for ESC and non-ESC vehicles.
Appendix A: Figures
Figure 1.
Primary Categories of Single-Vehicle Rollovers for ESC Vehicles
Figure 2.
Critical Events Following Road Departure Without Driver Input
Figure 3.
Vehicles Reentering Road After Departure Without Driver Input
Figure 4.
Driver Factors Contributing to Single-Vehicle Rollovers of ESC Vehicles
Figure 5A.
Driver Factors for Departure from Road without Driver Input (143)
Figure 5B.
Driver Factors for Loss of Directional Control (148)
Figure 6.
Comparisons of Roadway Conditions
Appendix B: Variables Coded for ESC Police Accident Report Analysis
Abrupt Elevation Change
Critical Event (up to 10)
Critical Event Text (up to 10)
Departed Road (up to 10)
Departed Road Text (up to 10)
Departure from Road (accident level)
Driver Factor Description (up to 4)
Divided Road
Driver Distraction Inattention
Driver Distraction Inattention Text
Grade of Road
Lighting Conditions
Loss of Control Event Description
Loss of Path Control (up to 10)
Loss of Path Control Basis (up to 10)
Number of Lanes in Travel Direction
Object Impacted (up to 10)
Object Impacted Text (up to 10)
Object Surface Tripped Upon
Object Surface Tripped Upon Text
Overcorrection During Accident
Pre Event Movement
Pre Event Movement Text
Quarter Roll Description
Quarter Rolls
Reenter Road
Report Number
Roadway Alignment
Roadway Surface
Roadway Surface Condition
Roadway Surface Condition Text
Rollover Initiation Location
Rollover Initiation Type
Rollover Initiation Type Text
Rollover Type
Shoulder Configuration
Shoulder Surface
Skidding During Accident
Speed Limit
Surface Condition
Time of Day
Vehicle Left Road Prior to Rollover
Vehicle Defect
Vehicle Pulling Trailer
Weather Conditions
Weather Conditions Text
Yawing During Accident
REFERENCES
- Aga M, Okada A. Analysis of Vehicle Stability Control’s Effectiveness from Accident Data. Toyota Motor Corporation; Japan: Toyota Techno Service Corporation; Japan: 2003. ESV Paper No. 541. [Google Scholar]
- Bahouth GT. Real World Crash Evaluation of Vehicle Stability Control Technology. PIRE. 2005 Sep; [PMC free article] [PubMed] [Google Scholar]
- Dang JN. Statistical Analysis of the Effectiveness of Electronic Stability Control (ESC) Systems. NHTSA Draft Technical Report, US Department of Transportation; Washington, D.C. Sept. 2006. [Google Scholar]
- Dang JN. Statistical Analysis of the Effectiveness of Electronic Stability Control (ESC) Systems – Final Report. National Highway Traffic Safety Administration, Final Report, No. DOT HS 810 794, July 2007.
- Farmer CM. Effect of Electronic Stability Control on Automobile Crash Risk. Study for Insurance Institute for Highway Safety; Arlington, Virginia: Oct, 2004. published in Traffic Injury Prevention Vol. 5, pp. 317-325. [DOI] [PubMed] [Google Scholar]
- Forkenbrock GJ, O’Harra BC, Elsasser D. An Experimental Evaluation of 26 Light Vehicles Using Test Maneuvers That May Induce On-Road, Untripped Rollover and a Discussion of NHTSA’s Refined Test Procedures—Phases VI and VII of NHTSA’s Light Vehicle Rollover Research Program. National Highway Traffic Safety Administration, Final Report, DOT HS 809 547, Oct. 2003.
- Green PE, Woodrooffe J. The Effectiveness of Electronic Stability Control on Motor Vehicle Crash Prevention. University of Michigan. Transportation Research Institute report, UMTRI; 2006–12 April 2006. [Google Scholar]
- Lie A, Tingvall C, Krafft M, Kullgren A. The Effectiveness of ESC (Electronic Stability Control) in Reducing Real Life Crashes and Injuries. 19th International Technical Conference on the Enhanced Safety of Vehicles (ESV); June 2005. [Google Scholar]
- Mercedes-Benz. Updated Accident Statistics: Mercedes Cars Have Fewer Accidents. ESP® press release, November 2006 [captured from Continental Teves, Inc. [January 22, 2007]. www.contitevesna.com/pdfs/MercedesStudyPressInfo.pdf.
- National Highway Transportation Safety Administration. Federal Motor Vehicle Safety Standards; Electronic Stability Control Systems; Controls and Displays. FMVSS 126, Final Rule, Docket No. NHTSA-2007-27662, April 6, 2007.
- Padmanaban J. Field Performance Study of Electronic Stability Control System Effectiveness in US Fatal Crashes. 2007 International IRCOBI Conference on the Biomechanics of Impact; Maastricht, Netherlands. September 19–21, 2007. [Google Scholar]
- Papelis Y, Brown T, Watson G, Holtz D, Pan W. Study of ESC-Assisted Driver Performance Using a Driving Simulator; University of Iowa. Report No. N04–003-PR; 2004. [Google Scholar]
- Thomas P, Frampton R. Real World Assessment of Relative Crash Involvement Rates of Cars Equipped with Electronic Stability Control. 20th International Technical Conference on the Enhanced Safety of Vehicles; Lyon, France. June 2007. Paper 07-0184. [Google Scholar]
- Viano DC, Parenteau C. Case Study of Vehicle Maneuvers Leading to Rollover: Need for a Vehicle Test Simulating Off-Road Excursions, Recovery and Handling. Society of Automotive Engineers. SAE 2003-01-0169; 2003. [Google Scholar]