INVESTIGATION OF SURROGATE ROAD SAFETY MEASURES AT INTERSECTIONS

Abstract

Intersections are one of the most crash-prone locations in a transportation network, with 33.2% of total crashes in India in 2018. Among 1,55,069 intersection crashes, unsignalized and signalized intersections accounted for 1,28,550 and 13,726 crashes causing 38,397 and 3,325 deaths, respectively. Unsignalized intersections in India behave like uncontrolled intersections as priority is not maintained by traffic and they have consistently proved to be very unsafe traffic facilities. At urban signalized intersections, the presence of dilemma zone (DZ) during the yellow interval results into red light violations (RLV) and sudden stopping of vehicles near the stop line. These two ill effects of DZ are the major factors contributing to motor vehicle crashes at signalized intersections. Traffic manoeuvres at intersection locations under heterogeneous traffic conditions are highly complex and potentially very unsafe. These have a direct influence on the occurrences of severe collision types such as right angle, right-turn right angle, right-turn head-on (left-hand driving) and rear-end crashes. Despite several efforts being made towards crash reduction using traditional crash data-based assessment, the number of road crashes, deaths, or injuries have been increasing over the years. In order to get rid of the problems associated with conventional methods in terms of crash data quality, long observation period and its reactive nature, the present study conducts proactive safety assessment at both types of intersections using a surrogate safety measure-based approach known as “Traffic Conflict Technique”. Consequently, this study develops different approaches to identify the unsafe real-time interactions (critical conflicts) between vehicles in terms of collision probability, criticality and severity. Later, the effects of several low-cost countermeasures on severe crash types are evaluated at both intersection types. Necessary data are collected from ten unsignalized and ten signalized intersections located in the National Capital Region (NCR) and Delhi respectively. To propose proximal indicators and validate proposed methods based on the crash-conflict relationship, these areas are specifically chosen. The initial site selection is done based on police recorded crash data where a substantial amount of right-angle, right turn related crashes and rear-end collisions are observed for the past 3-5 years. Later, geometric configurations and traffic conditions are utilized to make the final shortlisting of the study sites. Based upon the collision occurrence time, video data collection is done in the field to capture traffic movement related information.This study initially proposes a simpler and effective safety method to evaluate crossing conflicts which produce the most severe crash types (right turn related collisions) using a popular proximal indicator, Post Encroachment Time (PET). PET is estimated for conflict between through and right-turning traffic as a time gap when a right turner leaves the conflict zone and a through traffic enters into it. A PET threshold is determined using quantitative and qualitative approaches to identify critical crossing conflicts. In the first approach, relationships between right-turn related crashes and crossing conflicts are built with respect to each PET threshold for all conflicting vehicles as well as individual vehicle categories involved in conflicts. Later, a qualitative analysis is carried out by ranking the sites based on cumulative PET and related crash counts to verify the proposed quantitative technique. A PET threshold of 1 s is obtained from both the techniques and used further to identify critical conflicts for unsignalized intersections. However, at Indian unsignalized intersections, the conflicting speed of major road traffic is high, reflecting higher conflict criticality/severity, which can result in a severe collision. To identify the criticality of conflicts, a new speed parameter, critical speed, is introduced based on the stopping sight distance (SSD) criterion. A crossing conflict is designated as a critical one when the conflicting speed of through vehicle is observed to be more than the corresponding critical speed. Later, the appropriateness of the proposed indicator is verified by developing a crash-conflict relationship considering critical crossing conflicts and five years’ right turn related crash data. Another observation is that when major road traffic suddenly stops or decelerates due to the obstruction created by a non-prioritized violating vehicle coming from major/minor road, a secondary collision type, i.e., rear-end collision takes place. Consequently, the present study proposes a novel approach for identifying critical rear-end conflicts in terms of collision probability by utilizing two time-based proximal indicators, Time-to-Stop (Ts) and Modified Time to Collision (MTTC). The difference between these two indicators (Δt) is proposed as a new surrogate safety measure to evaluate rear-end conflicts. A 2 s threshold value of Δt is estimated from survival probability function to determine critical rear-end conflicts. To investigate the after-effects of probable severe crash types, the severities associated with critical crossing and rear-end conflicts are determined by proposing a conflict severity index (CSI). This index combines time-based indicators for different conflict types, the expected loss in kinetic energy after collision (ΔKE). ΔKE indicates the expected level of severity that affects the person riding/sitting inside the colliding vehicles if a collision takes place. Study results found that CSI increases with the increasing speed of right-of-way vehicle, conflict angle between cross-traffic and mass ratio between involved vehicles. Once the safety assessment is done based on field observations, microsimulation modelling is carried out in PTV VISSIM 8.0 platform. Utilizing the Surrogate Safety Assessment Model (SSAM), safety effects of different traffic calming measures and management techniques are evaluated at ten unsignalized intersections. These strategies include speed calming measures such as circular hump, trapezoidal hump, right-turn storage lane, and Median U-turn. For the study purpose, initially, the networks of one 3-legged and another 4-legged intersections have been prepared in VISSIM and calibrated with the 85th percentile accepted gap time. Median U-turn shows the maximum reduction for both crossing and rear-end conflicts. The performance of the right-turn storage lane is evaluated at three sites where the median width ranges from 4 to 4.5 m, showing the second-highest reduction in conflicts at these sites preceded by Median U-turn. Between two speed humps, trapezoidal hump is found to be a more effective measure by increasing safety at selected intersections. At signalized intersections, the effectiveness of the green signal countdown timer (GSCT) on the presence of Dilemma Zone (DZ) has been investigated and the length of both DZ I and II are found to be reduced when GSCT is present. Later, GSCT’s safety effects on RLVs and resulting right angle and right turn related crashes are also studied thoroughly. For this purpose, a before-after with comparison group method is used considering five intersections comprising of with and without timers from Delhi. At the with-timer sites (i.e., treatment sites where GSCTs are installed), several RLV characteristics up to initial 10 s of red such as RLV frequency, percentage of RLV cycles, RLV intensity, maximum violation time are observed to be less than comparison sites (without-timer sites). The percentage reduction in average hourly conflict counts at treatment sites due to the installation of GSCT is found as 59.47%. Later, an adjustment of yellow interval and provision of all-red clearance interval have been attempted as cost-effective countermeasures in microsimulation tool VISSIM. At GSCT enabled sites, the estimated yellow + all-red interval is found in the range of 5.62 s to 6.16 s which is higher than field allocated yellow timing (5s). Consequently, two timer-on and off models are prepared in VISSIM and calibrated as well as validated with average hourly outflow. A number of possible modifications of field conditions are attempted in VISSIM such as i) GSCT+ increase in yellow interval up to 6 s and ii) integration of separate yellow and all-red clearance at all the study sites. Among all countermeasures, a combination of GSCT + integration of separate yellow and all-red clearance shows the maximum benefit in reducing both crossing and rear-end conflicts. The study shows that GSCT works as an effective time-warning device at intersections which helps in reducing drivers’ dilemma as most of them utilize the end of green timer information for making better decisions of stopping/crossing and chances of severe crash types at intersections. On the other hand, the provision of appropriate yellow and all-red durations helps drivers to get rid of the dilemma regarding stopping/crossing during the yellow interval and complete the crossing during the all-red interval, respectively. This is the first comprehensive study in a developing country that has shed light on proactive safety evaluation using surrogate safety measure-based approach “Traffic Conflict Technique” at unsignalized and signalized intersections under mixed traffic conditions. Each of the proposed safety analysis methods is validated using a sound crash database. These can be used further for safety evaluation at intersections without the need for crash data. The proposed countermeasures in this study are verified to be a potential tool for reducing severe crashes at both intersection types. However, this study is limited to unsignalized intersections located on 4-lane divided highways and 4-legged signalized intersections from Delhi.