Introduction

This research extends the formulation of a new class of dynamic traffic flow model-hybrid models (Laval, 2004) &mdash and demonstrate that it explains and replicates existing empirical data on merge bottlenecks. Recent findings show that hybrid models are able to explain the effects on capacity of slow moving obstructions as observed in Mũnoz and Daganzo (2002), with only four observable parameters. This important finding could become a major contribution because it strongly suggests that the same physical principle can explain the elusive "capacity drop" phenomenon commonly observed at merge bottlenecks.

This research is divided in two stages. The first stage extends hybrid models to incorporate mandatory lane-changing, as the current formulation only accommodates for discretional lane-changing. The second stage includes the simulation of the experiments in Rudjanakanoknad (2004), which fingerprints an on-ramp bottleneck under time-varying metering rates. This data source is the first solid evidence showing that high system capacities can be restored by modulating the metering rates.

If we find that the proposed extension can replicate these observations &mdash as we confidently expect &mdash an important contribution to freeway operations would be made. In particular, novel ITS solutions that focus on the cause of traffic breakdown can now be designed, and efficient numerical simulation would be possible.

Research definition

Recent findings from our Department show that disruptive lane-changing (DLC) maneuvers are at the root of traffic instabilities. We call DLC a lane-changing maneuver that causes a moving obstruction because its initial speed is lower than the speed on the target lane. It was shown in Laval (2004) that a multi-lane theory based on DLC is able to explain and replicate the experimental evidence on the discharge rate of moving bottlenecks reported in Mũnoz and Daganzo (2002). In this latter reference, it was found that the lower the speed of the moving bottleneck, the lower the bottleneck discharge rate. This observation can be explained with DLC theory because the lower the speed of the moving bottleneck, the lower the speed of lane-changing vehicles in the queue, and hence the greater the impact on capacity.

It is expected that the same physical principle explain the puzzling behavior of merge bottlenecks, in particular the capacity drop phenomenon and how to prevent it. However, there is an important difference between these types of bottlenecks that is not taken into account in the current formulation of hybrid models. The lane changes caused by a moving bottleneck can be considered "discretional" as they seek to increase travel speed; lane changes from the ramp are "mandatory" as the vehicle has no choice but to change lanes. It becomes clear that the flow priorities will be very different in each case and that a mandatory lane-changing model (MLM) must be incorporated. These models give the flow proportions at the merge point; i.e., the probability that a vehicle crossing the merge point comes from the shoulder lane or from the on-ramp. Once the appropriate MLMs have been identified, the hybrid modeling framework can be applied unchanged. This is the result of conjecturing that the mechanisms inducing traffic collapse are independent of the cause that triggers the DLCs.

With all, the proposed research includes two major tasks:

1) Formulation and calibration of MLMs. The calibration will be carried out using the experimental data in Rudjanakanoknad (2004) and Ahn (2004), which provide high-fidelity counts based on video data.

2) Simulation of the experiments in Rudjanakanoknad (2004) and comparison of results. Good agreement is expected with little additional calibration.

References

JA Laval. Hybrid models of traffic flow: impacts of bounded vehicle accelerations. PhD thesis, Dept. of Civil Engineering, Univ. of California, Berkeley, 2004.

JC Mũnoz and CF Daganzo. Moving bottlenecks: a theory grounded on experimental observation. In M.A.P. Taylor, editor, 15th Int. Symp. on Transportation and Traffic Theory, pages 441-462, Pergamon-Elsevier, Oxford,U.K., 2002.

J Rudjanakanoknad. PhD thesis, Dept. of Civil Engineering, Univ. of California, Berkeley, 2004.

S Ahn. Growth of oscillations in queued traffic. PhD thesis, Dept. of Civil Engineering, Univ. of California, Berkeley, 2004.