Traffic Injury Prevention
ISSN: 1538-9588 (Print) 1538-957X (Online) Journal homepage: http://www.tandfonline.com/loi/gcpi20
Statistical Characteristics of Wrong-Way Driving Crashes on Illinois Freeways Huaguo Zhou, Jiguang Zhao, Mahdi Pour-Rouholamin & Priscilla A. Tobias To cite this article: Huaguo Zhou, Jiguang Zhao, Mahdi Pour-Rouholamin & Priscilla A. Tobias (2015) Statistical Characteristics of Wrong-Way Driving Crashes on Illinois Freeways, Traffic Injury Prevention, 16:8, 760-767, DOI: 10.1080/15389588.2015.1020421 To link to this article: http://dx.doi.org/10.1080/15389588.2015.1020421
View supplementary material
Accepted online: 20 Mar 2015.Published online: 20 Mar 2015.
Submit your article to this journal
Article views: 44
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=gcpi20 Download by: [University Of South Australia Library]
Date: 11 October 2015, At: 03:16
Traffic Injury Prevention (2015) 16, 760–767 C Taylor & Francis Group, LLC Copyright ISSN: 1538-9588 print / 1538-957X online DOI: 10.1080/15389588.2015.1020421
Statistical Characteristics of Wrong-Way Driving Crashes on Illinois Freeways HUAGUO ZHOU1, JIGUANG ZHAO2, MAHDI POUR-ROUHOLAMIN1, and PRISCILLA A. TOBIAS3 1
Department of Civil Engineering, Auburn University, Auburn, Alabama CH2M HILL Inc., Chicago, Illinois 3 Illinois Department of Transportation, Bureau of Safety Engineering, Springfield, Illinois
Downloaded by [University Of South Australia Library] at 03:16 11 October 2015
2
Received 16 October 2014, Accepted 14 February 2015
Objective: Driving the wrong way on freeways, namely wrong-way driving (WWD), has been found to be a major concern for more than 6 decades. The purpose of this study was to identify characteristics of this type of crash as well as to rank the locations/interchanges according to their vulnerability to WWD entries. Methods: The WWD crash data on Illinois freeways were statistically analyzed for a 6-year time period (2004 to 2009) from 3 aspects: crash, vehicle, and person. The temporal distributions, geographical distributions, roadway characteristics, and crash locations were analyzed for WWD crashes. The driver demographic information, physical condition, and injury severity were analyzed for wrongway drivers. The vehicle characteristics, vehicle operation, and collision results were analyzed for WWD vehicles. A method was brought about to identify wrong-way entry points that was then used to develop a relative-importance technique and rank different interchange types in terms of potential WWD incidents. Results: The findings revealed that a large proportion of WWD crashes occurred during the weekend from midnight to 5 a.m. Approximately 80% of WWD crashes were located in urban areas and nearly 70% of wrong-way vehicles were passenger cars. Approximately 58% of wrong-way drivers were driving under the influence (DUI). Of those, nearly 50% were confirmed to be impaired by alcohol, about 4% were impaired by drugs, and more than 3% had been drinking. The analysis of interchange ranking found that compressed diamond interchanges, single point diamond interchanges (SPDIs), partial cloverleaf interchanges, and freeway feeders had the highest wrong-way crash rates (wrong-way crashes per 100 interchanges per year). Conclusions: The findings of this study call for more attention to WWD crashes from different aspects such as driver age group, time of day, day of week, and DUI drivers. Based on the analysis results of WWD distance, the study explained why a 5-mile radius of WWD crash location should be studied for WWD fatal crashes with unknown entry points. Keywords: wrong-way driving, crashes, entry points, statistical characteristics, interchanges
Introduction Driving the wrong way on freeways has been a consistent traffic safety problem since the interstate system was opened in the 1950s. Herein, a wrong-way driver is considered to be traveling in the wrong direction on a physically separated motorway or is traveling in the opposite direction along a one-way street (Scaramuzza and Cavegn 2007; Zhou and Pour-Rouholamin 2014). Various characteristics of the roadway, especially exit ramp and crossroad intersections, can contribute to wrong-way driving (WWD) issues. Lack of/inappropriate signage and pavement marking as well as Associate Editor Matthew Reed oversaw the review of this article. Address correspondence to Huaguo Zhou, Department of Civil Engineering, Auburn University, 238 Harbert Engineering Center, Auburn, AL 36849-5337. E-mail: [email protected] Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/gcpi.
confusing geometric design of such intersections are among the contributing factors. Figure A1 (see online supplement) shows right- and wrong-way movements at a ramp–crossroad intersection of a partial cloverleaf interchange. This design with offset left turn lanes and 2 parallel medians, according to our study and crash history, is identified as a confusing, WWD-prone geometric design. Recent statistics show that about 350 people are killed each year nationwide due to WWD crashes based on the NHTSA’s Fatality Analysis Reporting System (NHTSA 2013). In Illinois, 217 freeway WWD crashes occurred from 2004 to 2009, resulting in 44 killed and 248 injured (Zhou et al. 2012). The Illinois Department of Transportation (IDOT) decided that an in-depth investigation of WWD crashes could provide a better understanding of these events. The purpose of this research was to review such severe crashes in depth, determine what contributing factors are most commonly involved, and generate ideas to consider in reducing the frequency and severity of these crashes.
Downloaded by [University Of South Australia Library] at 03:16 11 October 2015
Wrong-Way Driving in Illinois WWD crashes tend to be more severe and have a greater likelihood of resulting in death or injury when compared to other types of crashes. Past studies (Cooner et al. 2004a, 2004b; Copelan 1989) showed that although a very small percentage of overall traffic crashes were caused by WWD, a relatively large percentage of fatal crashes were. Drivers and passengers in both wrong-way and right-way vehicles can be killed in WWD crashes. For example, of the 49 fatal WWD crashes on the New Mexico interstate highway system between 1990 and 2004, 35 drivers and 11 passengers in the wrong-way vehicles were killed; 18 drivers and 15 passengers in vehicles traveling in the correct direction were killed as well (Lathrop et al. 2010). WWD crashes are more prevalent during non-daylight hours, particularly in the early morning. In Texas, the 6 hours from midnight to 6:00 a.m. were when 52% of all WWD crashes occurred; however, only 10.4% of overall freeway crashes occurred during that time period (Cooner et al. 2004a). Thus, wrong-way crashes were well overrepresented during the early morning hours. Past studies (Braam 2006; Cooner et al. 2004a; Copelan 1989; North Texas Tollway Authority 2009) have consistently indicated that WWD crashes occurred more frequently during the weekends. Monthly distribution of WWD crashes varies among different states (Braam 2006; Cooner and Ranft 2008) and countries (Institute of Traffic Accident Research and Data Analysis 2002) showing no consistent trend. Research conducted in both California (Copelan 1989) and Texas (Cooner et al. 2004a, 2004b) has found that urban areas have many more WWD crashes than rural areas. Studies in Texas also found that most of the wrong-way collisions occurred in the inside lane of the correct direction and at locations with left-side exit ramps or one-way streets that transitioned into freeway sections (Cooner et al. 2004a). A study in The Netherlands from 1983 to 1998 found that 79% of WWD crashes took place on the main line of freeway, 5% on merge/diverge lanes, and 17% on ramps (Institute for Road Safety Research 2009). The characteristics of wrong-way drivers, such as driver sobriety, age, and gender, have been discussed in past studies. A significant portion of WWD crashes on freeways was caused by motorists driving under the influence (DUI) of alcohol or drugs. Most past studies concluded that younger drivers and older drivers are overrepresented in the WWD crashes. Most of the crashes caused by drivers in the young and middle age range were brought about by inattention, whereas most crashes caused by drivers in the senior age range occurred because of some physical illnesses such as dementia or confusion. The overwhelming majority of WWD crashes involved male drivers, and most of the female drivers were in the young age groups (Institute of Traffic Accident Research and Data Analysis 2002). Past studies indicated that WWD crashes were random events and it was difficult to identify high-crash locations for improvements. However, this study found that some interchange types are more conducive to WWD than others. A new method to identify high-crash locations for field review and countermeasure developments was proposed and demonstrated. This method has been implemented by IDOT districts and validated as an effective way to develop site-specific countermeasures to mitigate WWD incidents and crashes.
761 Methods Data Collection The IDOT original crash data in text format was studied and used in the WWD crash data collection. This database contains 3 different subdatabases: crash file, vehicle file, and person file; each subdatabase contains various information relevant to WWD crashes. For instance, the crash file includes all of the information on the crash characteristics such as crash location, time of day, severity, etc. Altogether, 632 possible WWD crashes on freeways were identified from the total 2,387,877 crashes from 2004 to 2009 in Illinois. After reviewing the hardcopy crash reports of these 632 possible crashes, 217 WWD crashes were confirmed. Cross-median crashes are often reported as WWD crashes, which have been excluded after reviewing hardcopies of crash reports. Table 1 lists the number of crashes per year during this time. It should be noted that the overall total crashes significantly decreased in 2009 as a result of the Illinois crash reporting threshold changing from $500 to $1,500 in that year. Data Analysis The purpose of the data analysis was to investigate the statistical characteristics of WWD crashes, such as when and where they occurred and what the main contributing factors are. In this study, WWD crash data were analyzed from these 3 different perspectives based on the 3 subdatabases: crash, person, and vehicle. For each of these subdatabases, different characteristics were separately considered and divided into discrete categories for further analysis and to determine the role of that particular factor in the related subdatabase. These subdatabases are complementary and the information for the same crash can be linked together from the 3 different subdatabases based on the crash identification number.
Results Crash Characteristics In the crash subdatabase, each crash record consisted of 70 variables in either text or numerical format. From this database, the researchers determined the temporal distribution, geographic distribution, roadway characteristics, and other crash characteristics such as collision type and crash severity of WWD crashes. Temporal Distribution The temporal distributions of WWD crashes include crash year, month, day, and hour. The weather and light conditions were also considered part of the temporal distribution because both are highly correlated with the crash time (month and time of day). The annual WWD crash frequency from 2004 through 2009 varied between 31 and 40, with an average frequency of 36. The monthly distribution variation, from 9 to 25, was greater than the annual distribution. The crash data also indicated that about 42.8% of WWD crashes occurred on weekends per week and 51.2% between midnight and 5 a.m. per day. Figures A2 and A3 (see online supplement) illustrate the temporal distribution of WWD crashes.
762
Zhou et al.
Table 1. Number of crashes under different categories Year Category Total crashes Freeway crashes Possible WWD crashes Confirmed WWD crashes Percentage of freeway crashes Percentage of total crashes
2004
2005
2006
2007
2008
2009
Total
433,259 31,908 125 40 0.125 0.009
421,757 30,156 137 32 0.106 0.008
408,858 24,772 103 31 0.125 0.008
423,090 29,200 106 39 0.134 0.009
408,487 30,289 88 37 0.122 0.009
292,426 21,960 73 38 0.173 0.013
2,387,877 168,285 632 217 0.129 0.009
Downloaded by [University Of South Australia Library] at 03:16 11 October 2015
Approximately 80% of WWD crashes occurred when the road surface was dry, under clear weather conditions, and during nighttime hours. Geographical Distribution The geographical distribution characteristics of WWD crashes were extracted from the crash database, including county, city, and township. Although WWD crashes were reported in 43 of 102 counties in Illinois, about 64% of WWD crashes were located in the following 4 counties: Cook County (37.8%), St. Clair County (9.7%), Madison County (9.2%), and Will County (6.5%). It should be noted that Cook County and Will County are in the Chicago metropolitan area, and St. Clair County and Madison County are in the St. Louis metropolitan area. In 35 other counties, the reported WWD crashes averaged less than 4 per county during the 6-year period. Approximately 25.8% of the state’s wrong-way crashes were reported in the Chicago metropolitan area. Roadway Characteristics The crash data were then examined to determine the roadway characteristics of WWD crashes, specifically the route number, route type, traffic control device, road surface condition, road defects, intersection, and work zone. WWD crashes occurred on 30 different routes. Approximately 60% of them happened on 5 routes: I-55 (19.4%), I-94 (14.7%), I-57 (9.7%), I-74 (7.8%), and I-64 (7.4%). Ninety-five percent of WWD crashes occurred on roadways where no traffic control device malfunctions were reported. Furthermore, a construction/maintenance zone was noted in only approximately 7% of WWD crashes. Less than 3% of WWD crashes were related to a work zone, and 5% of WWD crashes were reported to be related to a specific interchange. Limited information on possible road design/traffic control problems was included in the crash reports, which makes it difficult for transportation engineers to develop engineering countermeasures at specific sites. Crash Locations The crash locations can be identified from the narrative description of the crash report hard copies for 56% of all wrongway crashes. The reported wrong-way crash location includes ramp (7%), left shoulder (3%), right shoulder (4%), and traveling lanes (86%). A large percentage of the wrong-way crashes on traveling lanes (59%) occurred on the most inside lane (lane 1).
Crash Characteristics The researchers reviewed crash characteristics, including the number of vehicles involved, collision type, and crash severity. Table A1 (see online supplement) summarizes the number of WWD crashes by type and severity in Illinois from 2004 to 2009. As can be seen, approximately 78.3% of WWD crashes were found to involve multiple vehicles and mostly resulted in head-on (45.6%) or sideswipe opposite-direction crashes (21.6%). The collisions for single-vehicle WWD crashes were mainly with fixed objects (14.3%). Altogether, the crash characteristics of multi- and single-vehicle crashes suggested that WWD crashes are more severe than other crash types. Most WWD crashes involved 2 or 3 vehicles, including the at-fault wrong-way vehicle, which collided with other vehicle(s) traveling in the correct direction(s). Findings indicated that the crash severity levels were directly related to collision types. Ninety-seven percent of fatal crashes were head-on crashes or opposite direction sideswipe crashes. The collision types for A-injury (incapacitating injuries) crashes were also mainly head-on (71.1%), opposite direction sideswipe (11.1%), and fixed object (8.9%). Almost 59.6% of head-on crashes caused fatalities or incapacitating crashes, whereas only 17% of opposite direction sideswipe crashes resulted in one or more fatalities and/or A-injuries. A noteworthy proportion (17.5%) of wrong-way vehicles fled the crash sites after the incident. The collision types were also related to the number of vehicles involved in WWD crashes. For example, rear-end crashes caused by wrong-way drivers were usually between 2 or more vehicles traveling in the correct directions that collided while avoiding wrong-way vehicles. Most of the rearend crashes were not severe. The collisions involving multiple vehicles were frequently either head-on crashes or oppositedirection sideswipe crashes. There were no fatal crashes for single-vehicle wrong-way crashes; however, many singlevehicle WWD crashes resulted in A- or B-injuries (nonincapacitating injuries). These statistics suggest that sending a warning message to right-way drivers using dynamic message signs can be helpful to reduce WWD crash severity by reducing the number of head-on crashes. Driver Characteristics To analyze the characteristics of wrong-way drivers, the person database for the 203 crash records was used. Logically, there should be a driver for each WWD crash; however, the difference between the number of drivers (203) and the number of WWD crashes (217) is because the information for 14 drivers could not be found in hardcopies (for example, the wrong-way
Wrong-Way Driving in Illinois
Downloaded by [University Of South Australia Library] at 03:16 11 October 2015
driver escaped from the scene). There were 32 variables in the person file, 11 of which were duplicates in different formats (code or text). A total of 660 persons were involved in the 217 WWD crashes, among which 6.7% were killed, 37.6% were injured, and 55.8% incurred no injuries. For those injured, 42.7 and 46.4% were A-injuries and B-injuries, respectively. Only about 10.9% of injuries were classified as possible injuries (Cinjury). Compared to other crash types, WWD crashes were more severe and large proportions resulted in severe injuries and/or fatalities. Driver Demographic Information Driver demographic information refers to date of birth, age, gender, and state of residence (as indicated on one’s driver’s license). The drivers were classified into different age groups based on the NHTSA criteria, which classify drivers under age 25 as young drivers and those 65 years and above as older drivers (NHTSA 1993). The results showed that older drivers were proportionally overrepresented in all crash types. With respect to the gender of wrong-way drivers, the database revealed that males represented nearly 67%, particularly those in the age groups of 21–24, 25–34, and greater than 65. Female wrong-way drivers were most prevalent in the 35–44 age groups. Demographic information indicated that most (76.8%) wrong-way drivers were licensed in the state of Illinois. The state with the next largest frequency (5.9%) was Missouri, which is not surprising given its proximity. Note that for roughly 16% of wrong-way crashes, the databases had no residence information for the drivers. Driver Physical Condition The driver’s physical conditions, including the apparent condition of the driver, the driver’s blood alcohol concentration (BAC) test result, and the driver’s vision, were analyzed to investigate the possible impact of DUI. The illegal BAC limits in Illinois are 0.00 for school bus drivers and drivers under the age of 21, 0.04 for commercial driver’s license holders, and 0.08 for drivers aged 21 and older. A large proportion of wrong-way drivers were found to be DUI: 49.8% by alcohol and nearly 4.4% by other drugs. However, the actual percentage should be higher because many drivers refused to take the test or were tested with no results. Eighty percent of the drivers completing a BAC test had a level greater than 0.10. Most of the DUI drivers were in the age range of 21–54, almost no senior wrong-way drivers were driving under the influence, and only about 16.7% of wrong-way drivers were in normal physical condition (not impaired). Driver vision was not reported as a possible reason for WWD crashes: nearly 99% of the reports note that the driver’s vision was not obscured or the information was unknown. Driver Injury Severity An analysis was conducted to identify the factors related to wrong-way driver injury severity. Figure A4 (see online supplement) shows the percentage of WWD crashes at different severity levels. Seat belt use, airbag deployment, driver age, gender, condition, and ejection or extrication were analyzed. More than 71% of wrong-way drivers were using their seat belts when WWD crashes occurred. Less than 7% of wrong-
763 way drivers who used a seat belt were killed in WWD crashes; however, the WWD fatality rate increased to more than 30% when seat belts were not used. Even though WWD crashes were more severe than most other crash types, almost 54.1% of wrong-way drivers who used a seat belt were not injured in the crashes. Airbag deployment for wrong-way vehicles was investigated as well. However, for more than half of wrongway vehicles, the airbag deployment was unknown. Less than 10% of wrong-way vehicles’ airbags were deployed from the front, side, or both. Additionally, the person database revealed that nearly 10% of wrong-way drivers were ejected or trapped/extricated as a result of the collision. Table A2 (see online supplement) illustrates the apparent relationship between driver crash severity level and driver condition. Approximately 80% of wrong-way drivers killed in the crashes were impaired by alcohol or drugs. Theoretically, there can be cases of combined impairments, but in the IDOT database, the variable “driver condition” could only be one of the conditions listed from Table A2. Therefore, based upon the information from the database, no such cases were observed. The DUI percentage was also relatively high among A-injured wrong-way drivers. The BAC for 65% of wrong-way drivers killed in the crashes was higher than 0.10, and 25% of them were completely ejected or trapped/extricated in the crashes. On the contrary, less than 10% of wrong-way drivers under normal physical conditions were killed in WWD crashes. The comparison between age, gender, and injury severity signified several trends. First, wrong-way drivers who were killed or incapacitated in WWD crashes were mainly in the age group of 21–45. Although the total number of fatalities and A-injuries contributed by older drivers was not as high as those of the 21–45 age group, the percentage of older wrong-way drivers who subsequently perish was much higher than the other age groups.
Vehicle Characteristics Vehicle characteristics including vehicle type, use, defects, commercial vehicle indicator, and number of occupants were analyzed. Results, which are summarized in Table A3 (see online supplement), indicate that no wrong-way vehicles were commercial vehicles. Nearly 68.5% of wrong-way vehicles were passenger cars. Approximately 90% of wrong-way vehicles were used for personal purposes and about 85.2% of wrongway vehicles had a single occupant. No vehicle defects were reported for more than 98% of WWD crashes. Vehicle Operation More than 80% of wrong-way vehicles collided with other vehicles and/or fixed objects with the front or the left/right front quarter panels, resulting in head-on collisions. Although information on the event, including location, was not recorded in the database for more than half of the WWD cases, the researchers reviewed the common events and locations for those available. The majority of known events for WWD crashes were either “motor vehicle in transport” or “ran off roadway.” The most frequently reported locations for the crash events were on roadway pavement.
764
Zhou et al.
Downloaded by [University Of South Australia Library] at 03:16 11 October 2015
Collision Results The vehicle database included information about the collision result. Researchers found that more than 80% of wrongway vehicles that crashed were towed away, and approximately 5.4% of wrong-way vehicles caught fire after the crash. The findings also indicated that no wrong-way vehicles spilled hazardous material during or after the crash. The number of vehicles involved in WWD crashes ranged from one to 6 (Table A4, see online supplement). More than 63% of WWD crashes involved 2 vehicles compared to 18.4% for one vehicle or 13.4% for 3 vehicles. Wrong-Way Entry Characteristics Based on the reported wrong-way entry points for the crashes studied, researchers can compare predicted entry points and the length of wrong-way travel. The correlation of these parameters is then used to propose a calibrated method for predicting wrong-way entry points for WWD crashes where little is known. Wrong-Way Entry Points Identifying wrong-way entry points can help develop proper countermeasures to combat WWD at a specific interchange area. However, information on wrong-way entry points was usually unavailable from the crash database. To obtain the information on wrong-way entry points, the narrative description of the crash report hardcopies were reviewed case by case, and the crash locations were examined using aerial photographs. Some vehicles began driving the wrong-way after they crossed the median, made a U-turn on the freeway, or tried to leave the freeway from an entrance ramp and these crashes were excluded from the analysis. For nearly 22% of real WWD crashes, wrong-way entry points were recorded in the crash report hardcopies. For crashes without recorded wrong-way entry points, the first and second possible wrongway entry points at interchanges were evaluated. These possible entry points would be the nearest first and second exit ramps on the freeway, if the vehicle is driven in the correct direction. When selecting the wrong-way collision site as the starting point, researchers determined the first and second wrong-way entry points by searching the first and second exit
Fig. 1. Cumulative distributions for WWD distance.
from the freeway when driving from the collision site in the correct direction (downstream). Based on an interview with the California Department of Transportation (Caltrans), a field review of 8 km (5 miles) radius of WWD crash site will be conducted if the wrong-way entry points for the crashes are unknown. The first and second possible entry points in this research are typically located within 8 km (5 miles) from the crash locations. As shown in Table 2, the total number of recorded entry points was 48, and the numbers of first and second possible entry points were 148 and 147, respectively. Concerning the recorded, first, second, and total wrong-way entry points, compressed diamond (25.0, 29.7, 29.9, 29.2%) and diamond interchanges (33.3, 26.4, 25.9, 27.1%) were the top 2 interchange types (Table 2). WWD Distance A comparison of driving distance between recorded and predicted entry points was conducted to see whether the characteristics of entry points recorded in crash reports had characteristics similar to those predicted by the researchers. The cumulative WWD distance distribution for recorded, first, and second entry points was plotted (Figure 1). Results indicated that WWD distances for recorded and estimated first entry points were very close, as 2 pertinent lines in the figure nearly match each other. For example, the average driving distance for WWD crashes was 1.93 km (1.20 miles) when an entry point was known/recorded and 2.48 km (1.54 miles) when it was estimated as the first upstream interchange. On the contrary, the average distance for the second estimated entry point was 5.79 km (3.60 miles), which is less than the 8-km (5-mile) study range based on the current practice by Caltrans.
Table 2. Interchange types for WWD crash entry points (2004–2009) First estimated entry point
Recorded
Second estimated entry point
Total WWD entry points
Interchange type
#
%
#
%
#
%
#
%
Compressed diamond Diamond Partial cloverleaf Cloverleaf Rest area Freeway feeder Modified diamond Semidirectional SPDI Trumpet Directional
12 16 5 3 1 5 3 0 1 0 2
25.0 33.3 10.4 6.3 2.1 10.4 6.3 0.0 2.1 0.0 4.2
44 39 28 12 9 4 4 3 2 2 1
29.7 26.4 18.9 8.1 6.1 2.7 2.7 2.0 1.4 1.4 0.7
44 38 23 12 6 6 4 5 3 4 2
29.9 25.9 15.6 8.2 4.1 4.1 2.7 3.4 2.0 2.7 1.4
100 93 56 27 16 15 11 8 6 6 5
29.2 27.1 16.3 7.9 4.7 4.4 3.2 2.3 1.7 1.7 1.5
Total
48
100.0
148
100.0
147
100.0
343
100.0
Downloaded by [University Of South Australia Library] at 03:16 11 October 2015
Wrong-Way Driving in Illinois The differences between the estimated and recorded entry points led the researchers to conduct further statistical analysis. The findings demonstrated that the WWD distance for the recorded and first estimated entry point were not significantly different at the 95% significance level. This result can be drawn from Figure 2 as the confidence intervals of recorded and first estimated entry points largely overlap. In other words, if there is a considerable overlap between 2 confidence intervals at 95% level, it is implied that the 2 studied variables are not significantly different (P value is greater than .05). On the contrary, the findings suggested that the second estimated entry point was significantly different from both other categories at the 95% level, because no overlap between the confidence intervals of these points and recorded and first estimated entry points can be spotted in Figure 2. When conducting safety audits for crashes with unknown entry points, these findings suggest that there is value in the evaluation of both the first and second estimated entry points. This explains why some states request a field review of 8-km (5-mile) radius of a WWD fatal crash because it covers most of the first and second possible WWD entry points. Interchange Ranking The recorded, first, and second estimated wrong-way entry points were used to rank the top locations based on the summation of weighted recorded, first, and second estimated wrong-way entry frequencies. The “weighting factors,” which imply the relative importance of entry points, were introduced and identified using the results of statistical method employed earlier. These factors were also discussed with safety engineers at the state Department of Transportation to get their feedback and comments. Mathematically said, as the mean WWD distance is the key difference between the recorded and estimated entry points, these values for first and second estimated entry points were compared to that of the recorded entry points and a weighting factor was then calculated and assigned to make the mean values the same. Accordingly, a weight of 1.0 was assigned to the recorded entry point and a weight of 0.70 was calculated for the first estimated entry point and 0.30 for the second estimated entry point. Applying the weight of the first and second estimated entry points to their corresponding mean WWD distances in Figure 2 produces the weighted mean distances depicted in Figure 3. Analysis of results showed that these new weighted mean distances for the
Fig. 2. 95% Confidence intervals of mean distances (km) between entry points and crash locations.
765
Fig. 3. Weighted confidence intervals of mean distances (km) between entry points and crash locations.
first and second estimated entry points are very close to the recorded ones. When considering the exposure (total number of each type of interchange in Illinois), WWD crash rate (total number of weighted wrong-way entries per 100 interchanges per year) can be calculated for all the interchange types using Eq. (1): C Ri nt =
100
2
i =0 wi Ei , Ni nt .T
(1)
where CRint is the WWD crash rate for each interchange type, wi is the assigned weight for specific type of entry point (1.00, 0.70, and 0.30), Ei is the number of specific types of entry points (Table 3), Nint is the total number of interchanges by type in Illinois (Table 3), and T is the study time period. For example, the WWD crash rate (CR) for compressed diamond interchange during the studied time period (2004–2009) is calculated as follows: CR =
100 × (1 × 12 + 0.70 × 44 + 0.30 × 44) = 16.7. (2) 56 × 6
All of the calculations are presented in Table 3 and interchanges are ranked using the calculated CR. Accordingly, compressed diamond, single point diamond interchange (SPDI), partial cloverleaf, and freeway feeder are the top 4 interchange types for the potential wrong-way entries. This method was applied to a statewide study in Illinois to recognize high-crash locations (specific location instead of specific type of interchange). Altogether there were 265 unique locations with 343 entries, of which 14% were recorded, 43.1% accounted for the first estimated, and 42.9% for the second estimated. Wrong-way entry was a sparse event, and the entry frequencies for most locations were relatively low, varying from one to 5. Among the 265 entry points, approximately 76% experienced one wrong-way entry and the remaining experienced 2 to 5 entries. Based on weighted wrong-way entry points, the top 10 locations were identified for field review. These locations included 4 compressed diamond, 3 partial cloverleaf interchanges, one diamond, one directional, and one SPDI. Comparing this list of interchanges with those in Table 3 indicates that including first and second likely entry points provides needed sensitivity to the selection process. For example, other studies implicated partial cloverleaf interchanges in WWD entries, but without the weighting process these would not have been selected in this case study. The field
766
Zhou et al.
Table 3. Interchange ranking by type using WWD crash rate Recorded (E 0 )
First estimated entry point (E 1 )
Second estimated entry point (E 2 )
Total no. of interchanges in Illinois (Nint )
Compressed diamond SPDI Partial cloverleaf Freeway feeder Cloverleaf Semidirectional Diamond Rest area Directional Modified diamond Trumpet
12 1 5 5 3 0 16 1 2 3 0
44 2 28 4 12 3 39 9 1 4 2
44 3 23 6 12 5 38 6 2 4 4
56 8 79 30 59 19 308 64 24 61 25
16.7 6.9 6.6 5.3 4.2 3.2 3.0 2.4 2.3 1.9 1.7
1 2 3 4 5 6 7 8 9 10 11
Total
48
148
147
733
4.4
—
Downloaded by [University Of South Australia Library] at 03:16 11 October 2015
Interchange type
reviews were conducted at the top 10 locations. During the field review, researchers identified improvements for wrongway related signage, pavement markings, and geometric elements. Site-specific countermeasures were developed and implemented by IDOT after this project. The results indicated that the method can effectively identify the high-crash locations for improvements. Six years of crash data from the IDOT were collected to identify WWD crashes. Out of 632 possible WWD crashes recognized from the crash database, 217 WWD crashes were verified by reviewing the hardcopies of those 632 crash reports. Statistical characteristics of WWD crashes on Illinois freeways were presented in this article as well as a method for identifying likely wrong-way entry points to the freeway system and, subsequently, rank the interchanges by type and location. The results indicated that approximately 42.9% of WWD crashes occurred on weekends per week and slightly over half (51.1%) of them from midnight to 5:00 a.m. per day. About 63% of WWD crashes were located in 4 counties adjacent to the urban areas of Chicago and East St. Louis, Illinois. Most (67.3%) of the multiple-vehicle WWD crashes were headon and sideswipe opposite-direction crashes. The road surface condition did not significantly contribute more to WWD crashes than other freeway crashes; however, a large portion of wrong-way drivers were DUI, among which roughly 50% were confirmed to be impaired by alcohol. These data confirm that the characteristics of wrong-way crashes continue to remain consistent, reinforcing findings from similar studies in other states and during previous decades. Because most WWD crashes (77.9%) have unknown originating locations, a new method was recommended to predict the first and second possible entry points. This method takes advantage of weighting factors given to each recorded/estimated entry point to identify the high-crash locations and interchange types for further field review. Accordingly, compressed diamond interchange, SPDI, and partial cloverleaf interchange were found to be the top 3 interchange type in terms of WWD crash rate. The findings and methods in this article have also been applied to a statewide study in Illinois to identify high-crash locations accurately and to develop site-specific and general countermeasures.
WWD crash rate (CRint )
Rank
This article summarizes the results of the research project’s phase I on identifying contributing factors regarding WWD on freeways. The second phase of this project is focused on identifying the wrong-way prevention countermeasures and finding empirical evidence to support the effectiveness of countermeasures used by DOTs. The results of the second phase of this project will be published in a separate technical paper.
Acknowledgment The authors would like to extend their sincere thanks to the Technical Review Panel (TRP) members for all their feedback and input.
Funding This study was sponsored and supported by the Illinois Department of Transportation (IDOT) and the Illinois Center for Transportation (ICT).
Supplemental Materials Supplemental data for this article can be accessed on the publisher’s website.
References Braam AC. Wrong-Way Crashes: Statewide Study of Wrong-Way Crashes on Freeways in North Carolina. Traffic Engineering and Safety System Branch, North Carolina Department of Transportation; 2006. Available at: https://connect.ncdot.gov/resources/safety/ Documents / Crash % 20Data % 20and % 20Information / WrongWay Crash.pdf. Accessed May 12, 2013. Cooner SA, Cothron AS, Ranft SE. Countermeasures for Wrong-Way Movement on Freeway: Guidelines and Recommendation Practices. College Station, TX: Texas Transportation Institute; 2004a. Cooner SA, Cothron AS, Ranft SE. Countermeasures for Wrong-Way Movement on Freeway: Overview of Project Activities and Findings. College Station, TX: Texas Transportation Institute; 2004b.
Downloaded by [University Of South Australia Library] at 03:16 11 October 2015
Wrong-Way Driving in Illinois Cooner SA, Ranft SE. Wrong-way driving on freeways: problems, issues and countermeasures. Paper presented at: Transportation Research Board 87th Annual Meeting; January 13–17, 2008; Washington, DC. Copelan JE. Prevention of Wrong-Way Accidents on Freeways. Sacramento, CA: California Department of Transportation; 1989. Institute for Road Safety Research. Fact Sheet: Wrong-Way Driving. 2009. Available at: http://www.swov.nl/rapport/Factsheets/ UK/FS Wrong way driving.pdf. Accessed May 3, 2013. Institute of Traffic Accident Research and Data Analysis. Highway Accidents Involving Dangerous Wrong-Way Traveling. 2002. Available at: http://www.itarda.or.jp/itardainfomation/english/info36/36top. html. Accessed April 23, 2013. Lathrop SL, Dick TB, Nolte KB. Fatal wrong-way collisions on New Mexico’s interstate highways, 1990–2004. J Forensic Sci. 2010;55:432–437. NHTSA. Addressing the Safety Issues Related to Younger and Older Drivers. Washington, DC: Author; 1993. DOT HS 807-957.
767 NHTSA. Fatality Analysis Reporting System (FARS) Encyclopedia. 2013. Available at: http://www-fars.nhtsa.dot.gov/main/index.aspx. Accessed July 5, 2013. North Texas Tollway Authority. Keeping NTTA Roadways Safe: WrongWay Driver Task Force Staff Analysis. 2009. Available at: https:// www.ntta.org/newsresources/safeinfo/wrongway/Documents/ WWDAnalysisAUG2011.pdf. Accessed May 7, 2013. Scaramuzza G, Cavegn M. Wrong-way drivers: extent-interventions. Paper presented at: European Transport Conference; October 17–19, 2007; Noordwijkerhout, The Netherlands. Zhou H, Pour-Rouholamin M. Guidelines for Reducing WrongWay Crashes on Freeways. Rantoul, IL: Illinois Center for Transportation/Illinois Department of Transportation; 2014. Zhou H, Zhao J, Fries R, et al. Investigation of Contributing Factors Regarding Wrong-Way Driving on Freeways. Rantoul, IL: Illinois Center for Transportation; 2012.
ISSN: 1538-9588 (Print) 1538-957X (Online) Journal homepage: http://www.tandfonline.com/loi/gcpi20
Statistical Characteristics of Wrong-Way Driving Crashes on Illinois Freeways Huaguo Zhou, Jiguang Zhao, Mahdi Pour-Rouholamin & Priscilla A. Tobias To cite this article: Huaguo Zhou, Jiguang Zhao, Mahdi Pour-Rouholamin & Priscilla A. Tobias (2015) Statistical Characteristics of Wrong-Way Driving Crashes on Illinois Freeways, Traffic Injury Prevention, 16:8, 760-767, DOI: 10.1080/15389588.2015.1020421 To link to this article: http://dx.doi.org/10.1080/15389588.2015.1020421
View supplementary material
Accepted online: 20 Mar 2015.Published online: 20 Mar 2015.
Submit your article to this journal
Article views: 44
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=gcpi20 Download by: [University Of South Australia Library]
Date: 11 October 2015, At: 03:16
Traffic Injury Prevention (2015) 16, 760–767 C Taylor & Francis Group, LLC Copyright ISSN: 1538-9588 print / 1538-957X online DOI: 10.1080/15389588.2015.1020421
Statistical Characteristics of Wrong-Way Driving Crashes on Illinois Freeways HUAGUO ZHOU1, JIGUANG ZHAO2, MAHDI POUR-ROUHOLAMIN1, and PRISCILLA A. TOBIAS3 1
Department of Civil Engineering, Auburn University, Auburn, Alabama CH2M HILL Inc., Chicago, Illinois 3 Illinois Department of Transportation, Bureau of Safety Engineering, Springfield, Illinois
Downloaded by [University Of South Australia Library] at 03:16 11 October 2015
2
Received 16 October 2014, Accepted 14 February 2015
Objective: Driving the wrong way on freeways, namely wrong-way driving (WWD), has been found to be a major concern for more than 6 decades. The purpose of this study was to identify characteristics of this type of crash as well as to rank the locations/interchanges according to their vulnerability to WWD entries. Methods: The WWD crash data on Illinois freeways were statistically analyzed for a 6-year time period (2004 to 2009) from 3 aspects: crash, vehicle, and person. The temporal distributions, geographical distributions, roadway characteristics, and crash locations were analyzed for WWD crashes. The driver demographic information, physical condition, and injury severity were analyzed for wrongway drivers. The vehicle characteristics, vehicle operation, and collision results were analyzed for WWD vehicles. A method was brought about to identify wrong-way entry points that was then used to develop a relative-importance technique and rank different interchange types in terms of potential WWD incidents. Results: The findings revealed that a large proportion of WWD crashes occurred during the weekend from midnight to 5 a.m. Approximately 80% of WWD crashes were located in urban areas and nearly 70% of wrong-way vehicles were passenger cars. Approximately 58% of wrong-way drivers were driving under the influence (DUI). Of those, nearly 50% were confirmed to be impaired by alcohol, about 4% were impaired by drugs, and more than 3% had been drinking. The analysis of interchange ranking found that compressed diamond interchanges, single point diamond interchanges (SPDIs), partial cloverleaf interchanges, and freeway feeders had the highest wrong-way crash rates (wrong-way crashes per 100 interchanges per year). Conclusions: The findings of this study call for more attention to WWD crashes from different aspects such as driver age group, time of day, day of week, and DUI drivers. Based on the analysis results of WWD distance, the study explained why a 5-mile radius of WWD crash location should be studied for WWD fatal crashes with unknown entry points. Keywords: wrong-way driving, crashes, entry points, statistical characteristics, interchanges
Introduction Driving the wrong way on freeways has been a consistent traffic safety problem since the interstate system was opened in the 1950s. Herein, a wrong-way driver is considered to be traveling in the wrong direction on a physically separated motorway or is traveling in the opposite direction along a one-way street (Scaramuzza and Cavegn 2007; Zhou and Pour-Rouholamin 2014). Various characteristics of the roadway, especially exit ramp and crossroad intersections, can contribute to wrong-way driving (WWD) issues. Lack of/inappropriate signage and pavement marking as well as Associate Editor Matthew Reed oversaw the review of this article. Address correspondence to Huaguo Zhou, Department of Civil Engineering, Auburn University, 238 Harbert Engineering Center, Auburn, AL 36849-5337. E-mail: [email protected] Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/gcpi.
confusing geometric design of such intersections are among the contributing factors. Figure A1 (see online supplement) shows right- and wrong-way movements at a ramp–crossroad intersection of a partial cloverleaf interchange. This design with offset left turn lanes and 2 parallel medians, according to our study and crash history, is identified as a confusing, WWD-prone geometric design. Recent statistics show that about 350 people are killed each year nationwide due to WWD crashes based on the NHTSA’s Fatality Analysis Reporting System (NHTSA 2013). In Illinois, 217 freeway WWD crashes occurred from 2004 to 2009, resulting in 44 killed and 248 injured (Zhou et al. 2012). The Illinois Department of Transportation (IDOT) decided that an in-depth investigation of WWD crashes could provide a better understanding of these events. The purpose of this research was to review such severe crashes in depth, determine what contributing factors are most commonly involved, and generate ideas to consider in reducing the frequency and severity of these crashes.
Downloaded by [University Of South Australia Library] at 03:16 11 October 2015
Wrong-Way Driving in Illinois WWD crashes tend to be more severe and have a greater likelihood of resulting in death or injury when compared to other types of crashes. Past studies (Cooner et al. 2004a, 2004b; Copelan 1989) showed that although a very small percentage of overall traffic crashes were caused by WWD, a relatively large percentage of fatal crashes were. Drivers and passengers in both wrong-way and right-way vehicles can be killed in WWD crashes. For example, of the 49 fatal WWD crashes on the New Mexico interstate highway system between 1990 and 2004, 35 drivers and 11 passengers in the wrong-way vehicles were killed; 18 drivers and 15 passengers in vehicles traveling in the correct direction were killed as well (Lathrop et al. 2010). WWD crashes are more prevalent during non-daylight hours, particularly in the early morning. In Texas, the 6 hours from midnight to 6:00 a.m. were when 52% of all WWD crashes occurred; however, only 10.4% of overall freeway crashes occurred during that time period (Cooner et al. 2004a). Thus, wrong-way crashes were well overrepresented during the early morning hours. Past studies (Braam 2006; Cooner et al. 2004a; Copelan 1989; North Texas Tollway Authority 2009) have consistently indicated that WWD crashes occurred more frequently during the weekends. Monthly distribution of WWD crashes varies among different states (Braam 2006; Cooner and Ranft 2008) and countries (Institute of Traffic Accident Research and Data Analysis 2002) showing no consistent trend. Research conducted in both California (Copelan 1989) and Texas (Cooner et al. 2004a, 2004b) has found that urban areas have many more WWD crashes than rural areas. Studies in Texas also found that most of the wrong-way collisions occurred in the inside lane of the correct direction and at locations with left-side exit ramps or one-way streets that transitioned into freeway sections (Cooner et al. 2004a). A study in The Netherlands from 1983 to 1998 found that 79% of WWD crashes took place on the main line of freeway, 5% on merge/diverge lanes, and 17% on ramps (Institute for Road Safety Research 2009). The characteristics of wrong-way drivers, such as driver sobriety, age, and gender, have been discussed in past studies. A significant portion of WWD crashes on freeways was caused by motorists driving under the influence (DUI) of alcohol or drugs. Most past studies concluded that younger drivers and older drivers are overrepresented in the WWD crashes. Most of the crashes caused by drivers in the young and middle age range were brought about by inattention, whereas most crashes caused by drivers in the senior age range occurred because of some physical illnesses such as dementia or confusion. The overwhelming majority of WWD crashes involved male drivers, and most of the female drivers were in the young age groups (Institute of Traffic Accident Research and Data Analysis 2002). Past studies indicated that WWD crashes were random events and it was difficult to identify high-crash locations for improvements. However, this study found that some interchange types are more conducive to WWD than others. A new method to identify high-crash locations for field review and countermeasure developments was proposed and demonstrated. This method has been implemented by IDOT districts and validated as an effective way to develop site-specific countermeasures to mitigate WWD incidents and crashes.
761 Methods Data Collection The IDOT original crash data in text format was studied and used in the WWD crash data collection. This database contains 3 different subdatabases: crash file, vehicle file, and person file; each subdatabase contains various information relevant to WWD crashes. For instance, the crash file includes all of the information on the crash characteristics such as crash location, time of day, severity, etc. Altogether, 632 possible WWD crashes on freeways were identified from the total 2,387,877 crashes from 2004 to 2009 in Illinois. After reviewing the hardcopy crash reports of these 632 possible crashes, 217 WWD crashes were confirmed. Cross-median crashes are often reported as WWD crashes, which have been excluded after reviewing hardcopies of crash reports. Table 1 lists the number of crashes per year during this time. It should be noted that the overall total crashes significantly decreased in 2009 as a result of the Illinois crash reporting threshold changing from $500 to $1,500 in that year. Data Analysis The purpose of the data analysis was to investigate the statistical characteristics of WWD crashes, such as when and where they occurred and what the main contributing factors are. In this study, WWD crash data were analyzed from these 3 different perspectives based on the 3 subdatabases: crash, person, and vehicle. For each of these subdatabases, different characteristics were separately considered and divided into discrete categories for further analysis and to determine the role of that particular factor in the related subdatabase. These subdatabases are complementary and the information for the same crash can be linked together from the 3 different subdatabases based on the crash identification number.
Results Crash Characteristics In the crash subdatabase, each crash record consisted of 70 variables in either text or numerical format. From this database, the researchers determined the temporal distribution, geographic distribution, roadway characteristics, and other crash characteristics such as collision type and crash severity of WWD crashes. Temporal Distribution The temporal distributions of WWD crashes include crash year, month, day, and hour. The weather and light conditions were also considered part of the temporal distribution because both are highly correlated with the crash time (month and time of day). The annual WWD crash frequency from 2004 through 2009 varied between 31 and 40, with an average frequency of 36. The monthly distribution variation, from 9 to 25, was greater than the annual distribution. The crash data also indicated that about 42.8% of WWD crashes occurred on weekends per week and 51.2% between midnight and 5 a.m. per day. Figures A2 and A3 (see online supplement) illustrate the temporal distribution of WWD crashes.
762
Zhou et al.
Table 1. Number of crashes under different categories Year Category Total crashes Freeway crashes Possible WWD crashes Confirmed WWD crashes Percentage of freeway crashes Percentage of total crashes
2004
2005
2006
2007
2008
2009
Total
433,259 31,908 125 40 0.125 0.009
421,757 30,156 137 32 0.106 0.008
408,858 24,772 103 31 0.125 0.008
423,090 29,200 106 39 0.134 0.009
408,487 30,289 88 37 0.122 0.009
292,426 21,960 73 38 0.173 0.013
2,387,877 168,285 632 217 0.129 0.009
Downloaded by [University Of South Australia Library] at 03:16 11 October 2015
Approximately 80% of WWD crashes occurred when the road surface was dry, under clear weather conditions, and during nighttime hours. Geographical Distribution The geographical distribution characteristics of WWD crashes were extracted from the crash database, including county, city, and township. Although WWD crashes were reported in 43 of 102 counties in Illinois, about 64% of WWD crashes were located in the following 4 counties: Cook County (37.8%), St. Clair County (9.7%), Madison County (9.2%), and Will County (6.5%). It should be noted that Cook County and Will County are in the Chicago metropolitan area, and St. Clair County and Madison County are in the St. Louis metropolitan area. In 35 other counties, the reported WWD crashes averaged less than 4 per county during the 6-year period. Approximately 25.8% of the state’s wrong-way crashes were reported in the Chicago metropolitan area. Roadway Characteristics The crash data were then examined to determine the roadway characteristics of WWD crashes, specifically the route number, route type, traffic control device, road surface condition, road defects, intersection, and work zone. WWD crashes occurred on 30 different routes. Approximately 60% of them happened on 5 routes: I-55 (19.4%), I-94 (14.7%), I-57 (9.7%), I-74 (7.8%), and I-64 (7.4%). Ninety-five percent of WWD crashes occurred on roadways where no traffic control device malfunctions were reported. Furthermore, a construction/maintenance zone was noted in only approximately 7% of WWD crashes. Less than 3% of WWD crashes were related to a work zone, and 5% of WWD crashes were reported to be related to a specific interchange. Limited information on possible road design/traffic control problems was included in the crash reports, which makes it difficult for transportation engineers to develop engineering countermeasures at specific sites. Crash Locations The crash locations can be identified from the narrative description of the crash report hard copies for 56% of all wrongway crashes. The reported wrong-way crash location includes ramp (7%), left shoulder (3%), right shoulder (4%), and traveling lanes (86%). A large percentage of the wrong-way crashes on traveling lanes (59%) occurred on the most inside lane (lane 1).
Crash Characteristics The researchers reviewed crash characteristics, including the number of vehicles involved, collision type, and crash severity. Table A1 (see online supplement) summarizes the number of WWD crashes by type and severity in Illinois from 2004 to 2009. As can be seen, approximately 78.3% of WWD crashes were found to involve multiple vehicles and mostly resulted in head-on (45.6%) or sideswipe opposite-direction crashes (21.6%). The collisions for single-vehicle WWD crashes were mainly with fixed objects (14.3%). Altogether, the crash characteristics of multi- and single-vehicle crashes suggested that WWD crashes are more severe than other crash types. Most WWD crashes involved 2 or 3 vehicles, including the at-fault wrong-way vehicle, which collided with other vehicle(s) traveling in the correct direction(s). Findings indicated that the crash severity levels were directly related to collision types. Ninety-seven percent of fatal crashes were head-on crashes or opposite direction sideswipe crashes. The collision types for A-injury (incapacitating injuries) crashes were also mainly head-on (71.1%), opposite direction sideswipe (11.1%), and fixed object (8.9%). Almost 59.6% of head-on crashes caused fatalities or incapacitating crashes, whereas only 17% of opposite direction sideswipe crashes resulted in one or more fatalities and/or A-injuries. A noteworthy proportion (17.5%) of wrong-way vehicles fled the crash sites after the incident. The collision types were also related to the number of vehicles involved in WWD crashes. For example, rear-end crashes caused by wrong-way drivers were usually between 2 or more vehicles traveling in the correct directions that collided while avoiding wrong-way vehicles. Most of the rearend crashes were not severe. The collisions involving multiple vehicles were frequently either head-on crashes or oppositedirection sideswipe crashes. There were no fatal crashes for single-vehicle wrong-way crashes; however, many singlevehicle WWD crashes resulted in A- or B-injuries (nonincapacitating injuries). These statistics suggest that sending a warning message to right-way drivers using dynamic message signs can be helpful to reduce WWD crash severity by reducing the number of head-on crashes. Driver Characteristics To analyze the characteristics of wrong-way drivers, the person database for the 203 crash records was used. Logically, there should be a driver for each WWD crash; however, the difference between the number of drivers (203) and the number of WWD crashes (217) is because the information for 14 drivers could not be found in hardcopies (for example, the wrong-way
Wrong-Way Driving in Illinois
Downloaded by [University Of South Australia Library] at 03:16 11 October 2015
driver escaped from the scene). There were 32 variables in the person file, 11 of which were duplicates in different formats (code or text). A total of 660 persons were involved in the 217 WWD crashes, among which 6.7% were killed, 37.6% were injured, and 55.8% incurred no injuries. For those injured, 42.7 and 46.4% were A-injuries and B-injuries, respectively. Only about 10.9% of injuries were classified as possible injuries (Cinjury). Compared to other crash types, WWD crashes were more severe and large proportions resulted in severe injuries and/or fatalities. Driver Demographic Information Driver demographic information refers to date of birth, age, gender, and state of residence (as indicated on one’s driver’s license). The drivers were classified into different age groups based on the NHTSA criteria, which classify drivers under age 25 as young drivers and those 65 years and above as older drivers (NHTSA 1993). The results showed that older drivers were proportionally overrepresented in all crash types. With respect to the gender of wrong-way drivers, the database revealed that males represented nearly 67%, particularly those in the age groups of 21–24, 25–34, and greater than 65. Female wrong-way drivers were most prevalent in the 35–44 age groups. Demographic information indicated that most (76.8%) wrong-way drivers were licensed in the state of Illinois. The state with the next largest frequency (5.9%) was Missouri, which is not surprising given its proximity. Note that for roughly 16% of wrong-way crashes, the databases had no residence information for the drivers. Driver Physical Condition The driver’s physical conditions, including the apparent condition of the driver, the driver’s blood alcohol concentration (BAC) test result, and the driver’s vision, were analyzed to investigate the possible impact of DUI. The illegal BAC limits in Illinois are 0.00 for school bus drivers and drivers under the age of 21, 0.04 for commercial driver’s license holders, and 0.08 for drivers aged 21 and older. A large proportion of wrong-way drivers were found to be DUI: 49.8% by alcohol and nearly 4.4% by other drugs. However, the actual percentage should be higher because many drivers refused to take the test or were tested with no results. Eighty percent of the drivers completing a BAC test had a level greater than 0.10. Most of the DUI drivers were in the age range of 21–54, almost no senior wrong-way drivers were driving under the influence, and only about 16.7% of wrong-way drivers were in normal physical condition (not impaired). Driver vision was not reported as a possible reason for WWD crashes: nearly 99% of the reports note that the driver’s vision was not obscured or the information was unknown. Driver Injury Severity An analysis was conducted to identify the factors related to wrong-way driver injury severity. Figure A4 (see online supplement) shows the percentage of WWD crashes at different severity levels. Seat belt use, airbag deployment, driver age, gender, condition, and ejection or extrication were analyzed. More than 71% of wrong-way drivers were using their seat belts when WWD crashes occurred. Less than 7% of wrong-
763 way drivers who used a seat belt were killed in WWD crashes; however, the WWD fatality rate increased to more than 30% when seat belts were not used. Even though WWD crashes were more severe than most other crash types, almost 54.1% of wrong-way drivers who used a seat belt were not injured in the crashes. Airbag deployment for wrong-way vehicles was investigated as well. However, for more than half of wrongway vehicles, the airbag deployment was unknown. Less than 10% of wrong-way vehicles’ airbags were deployed from the front, side, or both. Additionally, the person database revealed that nearly 10% of wrong-way drivers were ejected or trapped/extricated as a result of the collision. Table A2 (see online supplement) illustrates the apparent relationship between driver crash severity level and driver condition. Approximately 80% of wrong-way drivers killed in the crashes were impaired by alcohol or drugs. Theoretically, there can be cases of combined impairments, but in the IDOT database, the variable “driver condition” could only be one of the conditions listed from Table A2. Therefore, based upon the information from the database, no such cases were observed. The DUI percentage was also relatively high among A-injured wrong-way drivers. The BAC for 65% of wrong-way drivers killed in the crashes was higher than 0.10, and 25% of them were completely ejected or trapped/extricated in the crashes. On the contrary, less than 10% of wrong-way drivers under normal physical conditions were killed in WWD crashes. The comparison between age, gender, and injury severity signified several trends. First, wrong-way drivers who were killed or incapacitated in WWD crashes were mainly in the age group of 21–45. Although the total number of fatalities and A-injuries contributed by older drivers was not as high as those of the 21–45 age group, the percentage of older wrong-way drivers who subsequently perish was much higher than the other age groups.
Vehicle Characteristics Vehicle characteristics including vehicle type, use, defects, commercial vehicle indicator, and number of occupants were analyzed. Results, which are summarized in Table A3 (see online supplement), indicate that no wrong-way vehicles were commercial vehicles. Nearly 68.5% of wrong-way vehicles were passenger cars. Approximately 90% of wrong-way vehicles were used for personal purposes and about 85.2% of wrongway vehicles had a single occupant. No vehicle defects were reported for more than 98% of WWD crashes. Vehicle Operation More than 80% of wrong-way vehicles collided with other vehicles and/or fixed objects with the front or the left/right front quarter panels, resulting in head-on collisions. Although information on the event, including location, was not recorded in the database for more than half of the WWD cases, the researchers reviewed the common events and locations for those available. The majority of known events for WWD crashes were either “motor vehicle in transport” or “ran off roadway.” The most frequently reported locations for the crash events were on roadway pavement.
764
Zhou et al.
Downloaded by [University Of South Australia Library] at 03:16 11 October 2015
Collision Results The vehicle database included information about the collision result. Researchers found that more than 80% of wrongway vehicles that crashed were towed away, and approximately 5.4% of wrong-way vehicles caught fire after the crash. The findings also indicated that no wrong-way vehicles spilled hazardous material during or after the crash. The number of vehicles involved in WWD crashes ranged from one to 6 (Table A4, see online supplement). More than 63% of WWD crashes involved 2 vehicles compared to 18.4% for one vehicle or 13.4% for 3 vehicles. Wrong-Way Entry Characteristics Based on the reported wrong-way entry points for the crashes studied, researchers can compare predicted entry points and the length of wrong-way travel. The correlation of these parameters is then used to propose a calibrated method for predicting wrong-way entry points for WWD crashes where little is known. Wrong-Way Entry Points Identifying wrong-way entry points can help develop proper countermeasures to combat WWD at a specific interchange area. However, information on wrong-way entry points was usually unavailable from the crash database. To obtain the information on wrong-way entry points, the narrative description of the crash report hardcopies were reviewed case by case, and the crash locations were examined using aerial photographs. Some vehicles began driving the wrong-way after they crossed the median, made a U-turn on the freeway, or tried to leave the freeway from an entrance ramp and these crashes were excluded from the analysis. For nearly 22% of real WWD crashes, wrong-way entry points were recorded in the crash report hardcopies. For crashes without recorded wrong-way entry points, the first and second possible wrongway entry points at interchanges were evaluated. These possible entry points would be the nearest first and second exit ramps on the freeway, if the vehicle is driven in the correct direction. When selecting the wrong-way collision site as the starting point, researchers determined the first and second wrong-way entry points by searching the first and second exit
Fig. 1. Cumulative distributions for WWD distance.
from the freeway when driving from the collision site in the correct direction (downstream). Based on an interview with the California Department of Transportation (Caltrans), a field review of 8 km (5 miles) radius of WWD crash site will be conducted if the wrong-way entry points for the crashes are unknown. The first and second possible entry points in this research are typically located within 8 km (5 miles) from the crash locations. As shown in Table 2, the total number of recorded entry points was 48, and the numbers of first and second possible entry points were 148 and 147, respectively. Concerning the recorded, first, second, and total wrong-way entry points, compressed diamond (25.0, 29.7, 29.9, 29.2%) and diamond interchanges (33.3, 26.4, 25.9, 27.1%) were the top 2 interchange types (Table 2). WWD Distance A comparison of driving distance between recorded and predicted entry points was conducted to see whether the characteristics of entry points recorded in crash reports had characteristics similar to those predicted by the researchers. The cumulative WWD distance distribution for recorded, first, and second entry points was plotted (Figure 1). Results indicated that WWD distances for recorded and estimated first entry points were very close, as 2 pertinent lines in the figure nearly match each other. For example, the average driving distance for WWD crashes was 1.93 km (1.20 miles) when an entry point was known/recorded and 2.48 km (1.54 miles) when it was estimated as the first upstream interchange. On the contrary, the average distance for the second estimated entry point was 5.79 km (3.60 miles), which is less than the 8-km (5-mile) study range based on the current practice by Caltrans.
Table 2. Interchange types for WWD crash entry points (2004–2009) First estimated entry point
Recorded
Second estimated entry point
Total WWD entry points
Interchange type
#
%
#
%
#
%
#
%
Compressed diamond Diamond Partial cloverleaf Cloverleaf Rest area Freeway feeder Modified diamond Semidirectional SPDI Trumpet Directional
12 16 5 3 1 5 3 0 1 0 2
25.0 33.3 10.4 6.3 2.1 10.4 6.3 0.0 2.1 0.0 4.2
44 39 28 12 9 4 4 3 2 2 1
29.7 26.4 18.9 8.1 6.1 2.7 2.7 2.0 1.4 1.4 0.7
44 38 23 12 6 6 4 5 3 4 2
29.9 25.9 15.6 8.2 4.1 4.1 2.7 3.4 2.0 2.7 1.4
100 93 56 27 16 15 11 8 6 6 5
29.2 27.1 16.3 7.9 4.7 4.4 3.2 2.3 1.7 1.7 1.5
Total
48
100.0
148
100.0
147
100.0
343
100.0
Downloaded by [University Of South Australia Library] at 03:16 11 October 2015
Wrong-Way Driving in Illinois The differences between the estimated and recorded entry points led the researchers to conduct further statistical analysis. The findings demonstrated that the WWD distance for the recorded and first estimated entry point were not significantly different at the 95% significance level. This result can be drawn from Figure 2 as the confidence intervals of recorded and first estimated entry points largely overlap. In other words, if there is a considerable overlap between 2 confidence intervals at 95% level, it is implied that the 2 studied variables are not significantly different (P value is greater than .05). On the contrary, the findings suggested that the second estimated entry point was significantly different from both other categories at the 95% level, because no overlap between the confidence intervals of these points and recorded and first estimated entry points can be spotted in Figure 2. When conducting safety audits for crashes with unknown entry points, these findings suggest that there is value in the evaluation of both the first and second estimated entry points. This explains why some states request a field review of 8-km (5-mile) radius of a WWD fatal crash because it covers most of the first and second possible WWD entry points. Interchange Ranking The recorded, first, and second estimated wrong-way entry points were used to rank the top locations based on the summation of weighted recorded, first, and second estimated wrong-way entry frequencies. The “weighting factors,” which imply the relative importance of entry points, were introduced and identified using the results of statistical method employed earlier. These factors were also discussed with safety engineers at the state Department of Transportation to get their feedback and comments. Mathematically said, as the mean WWD distance is the key difference between the recorded and estimated entry points, these values for first and second estimated entry points were compared to that of the recorded entry points and a weighting factor was then calculated and assigned to make the mean values the same. Accordingly, a weight of 1.0 was assigned to the recorded entry point and a weight of 0.70 was calculated for the first estimated entry point and 0.30 for the second estimated entry point. Applying the weight of the first and second estimated entry points to their corresponding mean WWD distances in Figure 2 produces the weighted mean distances depicted in Figure 3. Analysis of results showed that these new weighted mean distances for the
Fig. 2. 95% Confidence intervals of mean distances (km) between entry points and crash locations.
765
Fig. 3. Weighted confidence intervals of mean distances (km) between entry points and crash locations.
first and second estimated entry points are very close to the recorded ones. When considering the exposure (total number of each type of interchange in Illinois), WWD crash rate (total number of weighted wrong-way entries per 100 interchanges per year) can be calculated for all the interchange types using Eq. (1): C Ri nt =
100
2
i =0 wi Ei , Ni nt .T
(1)
where CRint is the WWD crash rate for each interchange type, wi is the assigned weight for specific type of entry point (1.00, 0.70, and 0.30), Ei is the number of specific types of entry points (Table 3), Nint is the total number of interchanges by type in Illinois (Table 3), and T is the study time period. For example, the WWD crash rate (CR) for compressed diamond interchange during the studied time period (2004–2009) is calculated as follows: CR =
100 × (1 × 12 + 0.70 × 44 + 0.30 × 44) = 16.7. (2) 56 × 6
All of the calculations are presented in Table 3 and interchanges are ranked using the calculated CR. Accordingly, compressed diamond, single point diamond interchange (SPDI), partial cloverleaf, and freeway feeder are the top 4 interchange types for the potential wrong-way entries. This method was applied to a statewide study in Illinois to recognize high-crash locations (specific location instead of specific type of interchange). Altogether there were 265 unique locations with 343 entries, of which 14% were recorded, 43.1% accounted for the first estimated, and 42.9% for the second estimated. Wrong-way entry was a sparse event, and the entry frequencies for most locations were relatively low, varying from one to 5. Among the 265 entry points, approximately 76% experienced one wrong-way entry and the remaining experienced 2 to 5 entries. Based on weighted wrong-way entry points, the top 10 locations were identified for field review. These locations included 4 compressed diamond, 3 partial cloverleaf interchanges, one diamond, one directional, and one SPDI. Comparing this list of interchanges with those in Table 3 indicates that including first and second likely entry points provides needed sensitivity to the selection process. For example, other studies implicated partial cloverleaf interchanges in WWD entries, but without the weighting process these would not have been selected in this case study. The field
766
Zhou et al.
Table 3. Interchange ranking by type using WWD crash rate Recorded (E 0 )
First estimated entry point (E 1 )
Second estimated entry point (E 2 )
Total no. of interchanges in Illinois (Nint )
Compressed diamond SPDI Partial cloverleaf Freeway feeder Cloverleaf Semidirectional Diamond Rest area Directional Modified diamond Trumpet
12 1 5 5 3 0 16 1 2 3 0
44 2 28 4 12 3 39 9 1 4 2
44 3 23 6 12 5 38 6 2 4 4
56 8 79 30 59 19 308 64 24 61 25
16.7 6.9 6.6 5.3 4.2 3.2 3.0 2.4 2.3 1.9 1.7
1 2 3 4 5 6 7 8 9 10 11
Total
48
148
147
733
4.4
—
Downloaded by [University Of South Australia Library] at 03:16 11 October 2015
Interchange type
reviews were conducted at the top 10 locations. During the field review, researchers identified improvements for wrongway related signage, pavement markings, and geometric elements. Site-specific countermeasures were developed and implemented by IDOT after this project. The results indicated that the method can effectively identify the high-crash locations for improvements. Six years of crash data from the IDOT were collected to identify WWD crashes. Out of 632 possible WWD crashes recognized from the crash database, 217 WWD crashes were verified by reviewing the hardcopies of those 632 crash reports. Statistical characteristics of WWD crashes on Illinois freeways were presented in this article as well as a method for identifying likely wrong-way entry points to the freeway system and, subsequently, rank the interchanges by type and location. The results indicated that approximately 42.9% of WWD crashes occurred on weekends per week and slightly over half (51.1%) of them from midnight to 5:00 a.m. per day. About 63% of WWD crashes were located in 4 counties adjacent to the urban areas of Chicago and East St. Louis, Illinois. Most (67.3%) of the multiple-vehicle WWD crashes were headon and sideswipe opposite-direction crashes. The road surface condition did not significantly contribute more to WWD crashes than other freeway crashes; however, a large portion of wrong-way drivers were DUI, among which roughly 50% were confirmed to be impaired by alcohol. These data confirm that the characteristics of wrong-way crashes continue to remain consistent, reinforcing findings from similar studies in other states and during previous decades. Because most WWD crashes (77.9%) have unknown originating locations, a new method was recommended to predict the first and second possible entry points. This method takes advantage of weighting factors given to each recorded/estimated entry point to identify the high-crash locations and interchange types for further field review. Accordingly, compressed diamond interchange, SPDI, and partial cloverleaf interchange were found to be the top 3 interchange type in terms of WWD crash rate. The findings and methods in this article have also been applied to a statewide study in Illinois to identify high-crash locations accurately and to develop site-specific and general countermeasures.
WWD crash rate (CRint )
Rank
This article summarizes the results of the research project’s phase I on identifying contributing factors regarding WWD on freeways. The second phase of this project is focused on identifying the wrong-way prevention countermeasures and finding empirical evidence to support the effectiveness of countermeasures used by DOTs. The results of the second phase of this project will be published in a separate technical paper.
Acknowledgment The authors would like to extend their sincere thanks to the Technical Review Panel (TRP) members for all their feedback and input.
Funding This study was sponsored and supported by the Illinois Department of Transportation (IDOT) and the Illinois Center for Transportation (ICT).
Supplemental Materials Supplemental data for this article can be accessed on the publisher’s website.
References Braam AC. Wrong-Way Crashes: Statewide Study of Wrong-Way Crashes on Freeways in North Carolina. Traffic Engineering and Safety System Branch, North Carolina Department of Transportation; 2006. Available at: https://connect.ncdot.gov/resources/safety/ Documents / Crash % 20Data % 20and % 20Information / WrongWay Crash.pdf. Accessed May 12, 2013. Cooner SA, Cothron AS, Ranft SE. Countermeasures for Wrong-Way Movement on Freeway: Guidelines and Recommendation Practices. College Station, TX: Texas Transportation Institute; 2004a. Cooner SA, Cothron AS, Ranft SE. Countermeasures for Wrong-Way Movement on Freeway: Overview of Project Activities and Findings. College Station, TX: Texas Transportation Institute; 2004b.
Downloaded by [University Of South Australia Library] at 03:16 11 October 2015
Wrong-Way Driving in Illinois Cooner SA, Ranft SE. Wrong-way driving on freeways: problems, issues and countermeasures. Paper presented at: Transportation Research Board 87th Annual Meeting; January 13–17, 2008; Washington, DC. Copelan JE. Prevention of Wrong-Way Accidents on Freeways. Sacramento, CA: California Department of Transportation; 1989. Institute for Road Safety Research. Fact Sheet: Wrong-Way Driving. 2009. Available at: http://www.swov.nl/rapport/Factsheets/ UK/FS Wrong way driving.pdf. Accessed May 3, 2013. Institute of Traffic Accident Research and Data Analysis. Highway Accidents Involving Dangerous Wrong-Way Traveling. 2002. Available at: http://www.itarda.or.jp/itardainfomation/english/info36/36top. html. Accessed April 23, 2013. Lathrop SL, Dick TB, Nolte KB. Fatal wrong-way collisions on New Mexico’s interstate highways, 1990–2004. J Forensic Sci. 2010;55:432–437. NHTSA. Addressing the Safety Issues Related to Younger and Older Drivers. Washington, DC: Author; 1993. DOT HS 807-957.
767 NHTSA. Fatality Analysis Reporting System (FARS) Encyclopedia. 2013. Available at: http://www-fars.nhtsa.dot.gov/main/index.aspx. Accessed July 5, 2013. North Texas Tollway Authority. Keeping NTTA Roadways Safe: WrongWay Driver Task Force Staff Analysis. 2009. Available at: https:// www.ntta.org/newsresources/safeinfo/wrongway/Documents/ WWDAnalysisAUG2011.pdf. Accessed May 7, 2013. Scaramuzza G, Cavegn M. Wrong-way drivers: extent-interventions. Paper presented at: European Transport Conference; October 17–19, 2007; Noordwijkerhout, The Netherlands. Zhou H, Pour-Rouholamin M. Guidelines for Reducing WrongWay Crashes on Freeways. Rantoul, IL: Illinois Center for Transportation/Illinois Department of Transportation; 2014. Zhou H, Zhao J, Fries R, et al. Investigation of Contributing Factors Regarding Wrong-Way Driving on Freeways. Rantoul, IL: Illinois Center for Transportation; 2012.
