M-TRAIL|CEE|UMD - Traffic Safety

Traffic Safety

Image Source: FHWA

Crash Modification Factor Development

This research focuses on developing Crash Modification Factors (CMFs) for a range of left-turn signal phasing treatments at signalized intersections to better quantify their safety impacts. Given the high frequency and severity of left-turn-related crashes, especially those involving conflicting traffic and pedestrian movements, the study evaluates several key phasing conversions, such as transitioning from permissive to permissive-protected and from permissive-protected to fully protected left-turn phases. Using a cross-sectional analytical approach, the study integrates crash records, traffic volumes, roadway geometry, and surrounding land use characteristics to develop safety performance functions and estimate CMFs for each treatment. The results show that converting to fully protected phasing consistently reduces left-turn crash frequency, while permissive-protected treatments yield more variable safety outcomes. These findings provide evidence-based guidance to support more effective signal timing strategies and data-driven safety improvements at intersections.

Secondary Crash Identification

This research focuses on improving highway safety by developing a reliable method for identifying and analyzing secondary crashes, which are collisions that occur within the spatiotemporal influence of a preceding (primary) crash. Traditional crash databases often fail to accurately label these events, limiting the effectiveness of incident management strategies. To address this gap, the study proposes a hybrid identification method that combines static threshold rules with dynamic speed contour analysis to detect primary and secondary crash pairs from large-scale crash datasets. Using the identified crash records, a binary logit model was applied to uncover key contributing factors to secondary crash occurrence, such as snowy weather, rear-end collision types, multi-vehicle involvement, and adverse surface conditions. In parallel, the study utilized hierarchical ordered probit (HOPIT) models to assess crash injury severity patterns in both primary and secondary events. The findings offer critical insights into crash causation and severity, supporting the development of more effective incident response strategies and data-driven traffic safety interventions.

High Injury Network Development

As part of the SS4A initiative, M-TRAIL is collaborating with Baltimore City DOT to develop a comprehensive, data-driven framework for identifying and addressing roadway safety risks, particularly in high-injury corridors. The research leverages a rich, multi-layered dataset compiled at the street level, which includes detailed infrastructure attributes (e.g., speed limits, number of lanes, sidewalk presence, bike lanes, pavement condition), land use characteristics (e.g., proximity to schools, hospitals, and transit stops), and sociodemographic indicators (e.g., age distribution, income levels, car ownership, racial composition). In addition, high-resolution crash records and speed percentile data are used to map collisions and assess risk exposure. These data points are integrated to create a heatmap of crash severity and construct both high- and low-injury networks for comparative statistical analysis. The resulting high-injury network will serve as a qualitatively predictive tool to identify road segments at elevated risk for severe collisions, enabling the city to prioritize safety interventions.

Workzone Safety

We are advancing work zone safety and operational efficiency through the development of an intelligent Variable Speed Limit (VSL) control system tailored for freeway work zones. Our system integrates a macroscopic traffic flow model with a Kalman filter to predict and dynamically adjust speed limits based on real-time traffic conditions. By minimizing the difference between actual and ideal speed profiles, the model reduces abrupt deceleration and prevents the formation of traffic shockwaves—two major contributors to rear-end collisions in work zone areas. The VSL optimization framework is designed to smooth vehicle speed transitions, mitigate congestion buildup, and enhance safety without sacrificing traffic throughput. Through extensive simulations using a calibrated VISSIM environment, our approach has demonstrated significant improvements in both safety and efficiency metrics, such as reduced speed variance, lower travel times, and fewer vehicle stops. This work provides a scalable foundation for deploying proactive traffic management strategies that address the unique risks of lane closures and capacity disruptions in active work zones..

Page updated

Report abuse