1. Research Overview

1.1 Research Goals and Objectives

The goal of this research is to ultimately improve the safety of traffic operations at highway intersections in Utah. To achieve this goal, a number of objectives need to be met. This report represents the fulfillment of several, but not all, of these objectives. In effect, to improve the safety of traffic operations at intersections, the following objectives must be met:

1. Identify intersections at which crashes recur. Consider that some crashes are more severe than others, as well as crash frequencies, in making these identifications.
2. Using one or more measures of performance; "rank" intersections according to their crash statistics.
3. Select a subset of intersections that meet one or more performance criteria for further study.
4. For the selected intersections, determine the factors that may have contributed to incident occurrence.
5. Review and incorporate, as applicable, intersection safety information from the literature, the state of the practice in Utah, and the state of the practice elsewhere.
6. Diagnose intersection safety problems using Federal Highway Administration checklists and information from the literature.
7. Based on the results of the diagnoses, suggest pertinent countermeasures. Consider the effectiveness of the countermeasures as discussed in the literature.
8. Summarize the findings and recommend a draft statewide intersection safety plan.
9. Review, modify, and finalize the statewide intersection safety plan.
10. Implement the statewide intersection safety plan.
11. Monitor the effectiveness of the plan; modify and "tweak" it as needed.

This report emphasizes the first four objectives and partially fulfills the fifth and sixth objectives. Further study, subsequent to this research, is suggested to completely fulfill the fifth and sixth objectives, then to continue toward satisfying the seventh through ninth objectives. The tenth objective would require action beyond that of a research study, while the eleventh objective would constitute a post-implementation examination.

1.2 Research Scope, Approach and Limitations

This study considered all at-grade, roadway intersections with numbered state highways in Utah. Intersections between two highways not on the state route system were not considered. Intersections between a state route and a road that is not a state route were considered, as well as intersections between two state routes. As of 2003, Utah had 42,720 mi of road, of which 5,853 mi (13.7 percent) were on the state highway system (Highway Statistics 2003). There were 947 mi of Interstate and other freeways, all of which were under the state's jurisdiction. Hence, there were 4,906 mi of state routes having at-grade intersections; there were 41,773 mi of roads with intersections in the state, of which 11.7 percent were state routes. About 15,677 million vehicle-miles were traveled (VMT) on state routes in 2003, representing 43 percent of the total 36,390 million vehicle-miles traveled on all roads. There were 6,795 million VMT on state routes other than freeways in 2003, representing 24.7 percent of the state's 27,508 million VMT on non-freeway roads. Thus, if there is a direct, linear relationship between VMT and crashes, then one would expect 25 percent of all crashes in Utah to occur along state routes.

1.3 Crash Data Delivery System

Intersections were identified using the Intersection and SR (State Route) Intersection tools in the Utah Department of Transportation's (UDOT's) Crash Data Delivery System (CDDS). The CDDS is a web-delivered application capable of providing customized queries of the UDOT motor vehicle crash database. The capabilities of the CDDS, not all of which were used in this study, include information for decision-making on safety programs, safety-related performance measurement programs, and geographical information system (GIS) mapping for analysis. The CDDS was developed by a contractor, for UDOT, over a multiyear period starting in 2001. A second, enhanced version of the CDDS was available in a limited capacity toward the end of the contract for this study; the research team used several of the enhanced features. For example, a new "Points of Interest" tool allowed the team to efficiently double check on intersection locations and traffic control types. Upon selecting a route and range of milepoints, or a region or district, and the years of analysis, the tool delivered a list of all motor vehicle crashes occurring at intersections along the given route within the range of milepoints, or within the selected region or district. The scope of the list could be further limited by establishing a minimum number of crashes, or by varying the functional area of influence. Further information about the procedures that the research team used to extract information from the CDDS is provided in Section 8 of this report. The tools, along with the CDDS, were developed in separate efforts prior to the start of this study. Modifications and improvements to the tools continued, however, throughout the duration of this contract (at times, in fact, the tools were unavailable). The intersection tools were continuing to be advanced as of the preparation of this report. Subsequent research on intersection safety in Utah should benefit from these improved CDDS tools.

The research team found the CDDS to be quite useful, and essential to the type of study being conducted. For example, SR 186 was one of a handful of highways that were not in the CDDS intersection tools; crash data from this highway was, however, available from the CDDS' "Advanced Search" tool. To identify intersection-related crashes along SR 186, the research team used Traffic on Utah Highways 2003 to pinpoint major intersections and milepoints. Then, to identify "lesser" intersections, the team used a street map, scale, and measuring device. This tedious approach was used for SR 186, US 189 and US 191. While these activities were time-consuming, they revealed the efficiency of the CDDS in identifying intersection locations along numerous other major highways. The research team estimated that, without the CDDS, the study would have required triple the amount of time; or, the team would have accomplished about one-third the amount of work. Further, the research team was able to maintain a level of consistency and accuracy in its database searches. It was easy, for example, to transfer responsibilities between members of the research team, as well as to check each other's work. The CDDS querying capabilities enabled, for example, the rapid identification of intersections at which fatal or incapacitating injury crashes occurred by region or district. Although the research team made only limited use of the enhanced CDDS - because of the short amount of time - it was evident that the new tools and parameters (such as crash rates and an large number of intersection types) would be of even greater use to future highway safety analysis than the original CDDS. Further discussion of the research team's use of the CDDS is found in Section 8.

1.4 Definition of an Intersection

An intersection is a crossing or meeting of two or more roads, at grade. An intersection may be controlled by a traffic signal, stop signs or yield signs, or it may be uncontrolled. Where two or more roads meet or cross, the intersection may be "controlled" by a traffic circle or rotary. All types of controlled and uncontrolled intersections were considered in this research; the only limitation, as mentioned earlier, was that at least one road was a state route.

The influence area of an intersection extends beyond the boundaries of the physical area of an intersection. Stover (1996) found that all intersections have downstream and upstream functional areas. Upstream of an intersection, motorists perceive and react to downstream events, such as an upcoming stop sign or a changing traffic signal. Motorists also decelerate and maneuver into turn lanes and storage queues. Downstream of an intersection, drivers accelerate, make left and right turns, encounter left- and right-turning vehicles from the cross-street, and prepare for deceleration at locations that are farther downstream. The functional areas of the downstream activities are not as well-defined as the areas corresponding to the upstream activities. The lengths of the functional areas vary according to the prevailing speed of travel or speed limit. Any collisions that occur during the upstream or downstream activities are incorporated into the functional area of the intersection. Table 1.1 summarizes the pertinent values.

Table 1.1 Intersection Functional Areas (Distances) (from Stover 1996)

Notes: Speed is in mph. Distances are in ft. PIEV = perception-identification-emotion-volition time. The "desirable" PIEV is 2.0 sec, while the limiting PIEV is 1.0 sec. Braking rates are assumed to be 3.5 ft/sec² under desirable conditions and 4.5 ft/sec² under limiting conditions. SSD = stopping sight distance.

Application of the values in Table 1 is not straightforward. The strictest interpretation of the values would result in a set of two or more influence distances for each intersection, varying according to the motorist's direction and the speed limit on the given street. Such a detailed approach would be appropriate for the examination of specific intersections. For a general statewide analysis, however, it is most convenient to use an average value that can reasonably be applied to all intersections. An influence distance of 500 ft (actually, 0.09 mi or 475 ft) was selected for this study. This distance corresponds, roughly, to approach speeds of 40 mph. The 500-ft distance overestimates the influence area for intersections with approach speed limits less than 40 mph, and underestimates that for intersections with approach speeds greater than 40 mph. In cases of closely-spaced intersections, it can be difficult to isolate the critical intersection. For the purposes of this study, when intersections were closely spaced, the critical intersection was considered to be the busiest one (i.e., the greater number of entering vehicles) or, in some cases, the one with the greatest number of collisions.

1.5 Identification of Hazardous Intersections

There are several ways to identify hazardous intersections. Kononov (2002) noted that crash frequency and severity are commonly used to measure safety performance. Persaud et al. (2001), for example, used all, all injury, incapacitating injury, and fatal crashes to evaluate the effectiveness of roundabout conversions. A crash prediction model will typically use crash frequency as the dependent variable (Belanger, 1994; Bonneson, 2002). Numerous "refined" measures of intersection safety performance have been used, including the deviation from an average or expected number of crashes (Kononov, 2002; Kononov and Allen, 2003), accident or crash rates (Abo-Qudais and Al-Mughrabi, 2004; Elvik, 2004), conflict rates (Salman and Al-Maita, 1995; Sayed and Zein, 1999), and the number of "primary" and "secondary" conflicts (Katamine, 2000). The research team, with concurrence from the project's technical advisory committee (TAC), opted for three measures of intersection safety performance: the total number of crashes, the crash severity score (described in Section 2), and the crash rate. For the purposes of this research, the crash rate is defined as the ratio of the total number of crashes occurring during a given study period to the total number of vehicles entering the intersection during that period.

1.6 Intersection Crash Contributing Factors

Hauer et al. (2002) listed the following potential contributing factors for various collision types:

• Bicyclist not on shoulder or bike lane; no bicycling facilities (bicycle-vehicle crashes)
• Conflicting decisions between drivers moving in the same direction once a traffic signal changes to amber (rear-end crashes)
• Conflicts with right turn on red vehicles (left-turn crashes, pedestrian-vehicle crashes)
• Conflicts with crossing pedestrians or bicyclists (left-turn crashes, right-turn crashes)
• Delayed or slow driver perception and reaction; inattentiveness (all crashes)
• Impairment (all crashes)
• Lack of conspicuity (pedestrian-vehicle crashes, bicycle-vehicle crashes)
• Misjudgment of the speed of oncoming vehicles (left-turn crashes)
• Pedestrian crossing outside of designated crossing area (pedestrian-vehicle crashes)
• Poor sight distance (all crashes)
• Poor visibility (all crashes)
• Red light violations; noncompliance with signal (right-angle collisions, pedestrian-vehicle crashes, bicycle-vehicle crashes)
• Sudden lane changes (side-swipe incidents)
• Speed differences between vehicles moving in the same direction (rear-end crashes)
• Speeding (single vehicle incidents)
• Swerving to avoid a vehicle, other highway user, animal, or object (single vehicle incidents)

The preceding factors may be considered as fundamental to the resultant crash. However, it may take some sleuthing and crash reconstruction expertise to identify the true contributing factor in a given crash. Hauer et al. (2002) indicated some secondary contributing factors that may lead to or induce the primary factor, thereby resulting in a crash. The secondary factors include:

• Actuated traffic signal that changes too frequently.
• Adjacent land uses contributing (school, senior center, commercial district, park, bar, etc.)
• Bus stop adjacent a pedestrian crossing that is heavily used.
• Bus stop, near-side or far-side, too close to the intersection; no bus bay.
• Bus stop not adjacent a pedestrian crossing.
• Conflicts between left-turning and right-turn-on-red vehicles.
• Conflicts between U-turning and right-turn-on-red vehicles.
• Construction or highway maintenance lane closures or restrictions.
• Cross-sectional design issues (lane widths, shoulder widths, pavement edge drops, cross slopes).
• Crossing pedestrians, possibly jaywalking.
• Dedicated turning lane with no or inadequate upstream warning to highway users.
• Driveways located too close to an intersection.
• Heavy pedestrian crossing volumes.
• High operating speeds on one or more intersection approaches.
• Inadequate traffic signal clearance time interval.
• Intersection located at the base of a downgrade from one or more approaches.
• Intersection located downstream of a horizontal curve along one or more approaches.
• Intersection located downstream of a vertical curve along one or more approaches.
• Isolated traffic signal on a high-speed road; lack of long-distance, advanced detection.
• Lane, stop or crosswalk markings or symbols not visible.
• Lane terminates at or immediately downstream of an intersection.
• Limited turning radii at an intersection.
• Malfunctioning traffic signal(s).
• No bicycling facilities (bike lane, designated bike route, shoulder).
• No pedestrian facilities (pedestrian signals, pushbuttons, grade-separated crossing, median).
• Parallel or angle parking located close to an intersection.
• Pavement surface hazards.
• Permitted left turn phasing with limited opportunities to complete left turns.
• Permitted right turns with limited opportunities for turning; inadequate storage.
• Poor lane delineation or channeling.
• Poor nighttime visibility.
• Poor pavement surface friction.
• Poor sight distance for right turning drivers.
• Poor or reduced traffic signal visibility.
• Right turn lane with no or an inadequate downstream acceleration lane.
• Right turn lane with no or an inadequate upstream deceleration lane.
• Speed limit changes near an intersection.
• Stop line not positioned properly.
• Turning lane (left or right) with an inadequate storage length.
• Uncoordinated, closely-spaced traffic signals, resulting in frequent starts and stops.
• Unfamiliar drivers; inconspicuous street name or highway route signs.
• Unusual intersection layout.
• Wide intersection with long pedestrian crossing distances; inadequate ped signal time.
• Winter roadway surface issues.

Once the primary and/or secondary contributing factors are identified, mitigating strategies can be selected. Identification of the factors would, in general, need to be made on a case-by-case basis; that is, at the site-specific level. An investigation at the statewide, regional or district level, as in this study, can make only general observations regarding intersection improvements. Importantly, a statewide, regional, or district study such as this can indicate general needs, reveal the extent of safety problems, and provide guidance for planning, budgeting, and prioritization.