Modeling Speed Mean and Variance for Different Enforcement Conditions on Multilane Highways

Abstract

Two major characteristics that can impact the frequency and severity of traffic crashes are average speed and speed variance. This paper investigated these characteristics on 49 multilane highway segments in Jordan and developed prediction models for three different conditions: free (no enforcement), camera, and police enforcement. Speed data were collected during off-peak periods with fair weather conditions. Multiple linear regressions and nonlinear regressions were used to develop speed variance and average speed models. For the free condition, speed variance mainly depends on the difference between design speed and speed limit (DS-SL), with the speed variance decreasing as the difference reduces, while the average speed was positively correlated with the design speed. Camera enforcement causes speed variance and average speed to be positively correlated and linearly dependent with the speed limit. A quadratic relationship was found between speed variance during police enforcement and DS-SL, while the average speed during police enforcement depended on the speed limit and design speed. These modeling results, along with analysis of the collected data, can help operating agencies and roadway designers determine how various enforcement strategies affect average speed and speed variance and better set speed limits on existing and future roadways to improve safety.