Observations on Traffic Behavior in Freeway Weaving Bottlenecks: Empirical Study and Theoretical Modeling

Though there have been numerous studies of freeway weaving sections (i.e., segments in which an on-ramp is followed by an off-ramp), there remains a significant lack of empirical and theoretical understanding of the traffic behavior that causes weaving sections to become bottlenecks with varying discharge flows. The present research entails empirical analysis and theoretical modeling of what triggered the bottleneck activations and discharge flow changes in two freeway weaving sections. Both sites were recurrent bottlenecks during the rush with average discharge flow reduction from 9900 vph to 8600 vph, and investigations revealed that changes in the spatial patterns of vehicular lane-changes, especially among Freeway-to-Ramp (F-R) maneuvers, triggered these bottlenecks and caused variations in their discharge flows. When the F-R maneuvers were concentrated near a weaving section's on-ramp, they became more disruptive, resulting in bottleneck activations with diminished discharge flows. Findings further indicate that the spatial distributions of these lane changes, in turn, were dictated by the traffic conditions in the auxiliary lane (i.e., the lane connecting the off-ramp to the upstream on-ramp). Reductions in on-ramp flows increased the attractiveness of the auxiliary lane, thus motivating F-R drivers to perform their maneuvers nearer the on-ramp. Conversely, increases in on-ramp flows motivated F-R drivers to perform their maneuvers over a wider stretch of the weaving section. Based on these empirical findings, the study formulated a theory for mandatory lane changing (i.e., lane changes required of Origin-Destination pairings); and used this theory to enhance an existing microsimulation model of car-following and lane changing. With this enhanced theory, the driver's decision to attempt a lane change is determined by the vehicle's distance from the downstream end of the weaving section's diverge area, the number of lanes to be crossed in reaching the desired destination, and the difference in densities between the driver's target lane and her current one. The model reproduces the observed mechanisms of bottleneck activations and discharge flow changes in weaving sections. These empirical findings, together with the outcomes of simulation, point to two key features of driver behavior in weaving sections: (i) traffic conditions (especially densities) in an auxiliary lane influence drivers' decisions regarding where to perform mandatory lane changes; and (ii) the spatial distributions of these lane changes determine whether a weave section becomes a bottleneck and the discharge flows that would result when it does.