Safety Performance Evaluation of Integrated Active Traffic Management on Freeways Using Crash Frequency Models

<jats:p>This study evaluated the safety performance of an integrated Active Traffic Management system combining Variable Speed Limits, Lane Control Signs, and Part Time Shoulder Use on Central Florida freeways. Freeway segments operating under the integrated system were compared to similar untreated segments, matched on traffic and geometric characteristics, to evaluate total, rear-end, sideswipe, and injury crashes. Crash rates per million vehicle miles traveled were compared between treated and comparison segments, and crash frequency models were estimated to control for traffic exposure, segment configuration, entry and exit presence, and shoulder width. Models confirmed that treated segments experienced significantly lower crash frequencies than comparison segments for total, rear-end, sideswipe, and injury crashes. The Poisson Lognormal model provided the best fit for total crashes and estimated a crash modification factor of 0.596, representing a 40.4 percent reduction in expected total crash frequency associated with the integrated system. Because rear-end and sideswipe crashes were positively related to injury crashes, joint Poisson Lognormal models were also estimated. The joint models produced crash modification factors of 0.609 for rear-end crashes, 0.679 for sideswipe crashes, and 0.463 for injury crashes, corresponding to reductions of 39.1 percent, 32.1 percent, and 53.7 percent, respectively. Model also showed a stronger treatment effect at complex segments (merge, diverge, and weaving) than basic segments, consistent with the crash rate trend. These findings indicate that the integrated system improved freeway safety by helping drivers respond more smoothly to congestion, lane closures, and changing traffic conditions. The study provides practical evidence to support future Active Traffic Management planning and operational decisions.</jats:p>