Evaluation of the effectiveness of freeway traffic management systems using flow-occupancy diagrams and capacity analysis

Abstract

[ACCESS RESTRICTED TO THE UNIVERSITY OF MISSOURI AT AUTHOR'S REQUEST.] Over the past decades environmental sustainability and mitigating congestion has been a significant challenge in urban areas in many countries. Freeway traffic management systems are aimed to improve traffic conditions by better managing existing freeway facilities. This dissertation presents an evaluation methodology that consists of series of statistical tests to identify the changes in flow-occupancy plot and flow breakdown on freeways. Two-dimensional Kolmogorov-Smirnov test was used to examine the statistical significance of changes in the overall flow-occupancy distribution. Curve-fitting techniques were used to produce before and after flow-occupancy fitted plots. In order to compare the fitted plots, statistical tests were presented to identify if slopes of fitted lines were statistically different. One challenge in creating traffic flow plots is the identification of critical occupancy. An automated procedure based on clustering techniques was presented to alleviate the modeler discretion from identifying the critical occupancy value. The impact of freeway traffic management systems on the probability of flow breakdown was also investigated. A non-parametric product-limit method and parametric Weibull distributions were used to model flow breakdown distributions. The applicability of presented methodologies for evaluating the freeway traffic management systems was illustrated using two case studies; a variable speed limit system in St. Louis, Missouri, and a freeway ramp metering system in the Twin Cities, Minnesota. The two-dimensional Kolmogorov-Smirnov test was illustrated to be useful in identifying the changes in before and after study of flow-occupancy distributions. Clustering techniques were successfully applied to partition traffic data to two regimes and to create flow-occupancy diagrams and to identify flow breakdown. Finally, an examination of the breakdown probability distribution highlighted its values as a more complete performance measure than capacity or critical occupancy.