A Stochastic Modeling of Traffic Breakdown for Freeway Merge Bottlenecks and Variable Speed Limit Control Strategies Using Connected Automated Vehicle

This thesis develops a novel breakdown probability model based on microscopic driver behavior for a freeway merge bottleneck. Extending Newell's car following model, two elements of breakdown, trigger and propagation, are derived in terms of vehicle headway. A general breakdown probability is derived in terms of various parameters related to driver behavior and traffic conditions that can be treated as constants or stochastic with probability distributions. The proposed model is validated with real data. It was found the theoretical breakdown probability distribution accords well with the empirical counterpart within reasonable ranges of parameter values. Developed model suggests the breakdown probability (i) increases with flow and the merging spacing, (ii) decreases with the merging speed and aggressive driver characteristics, and interestingly, (iii) increases with the deviation in headway. To achieve the optimum traffic state, the headway adjustment area is suggested as a proactive control with a combination of connected automated vehicle (CAV) and VMS. In this area, random headways evolve to more uniform headways without inducing controlled congestion propagating upstream.;Reactive VSL control is also developed in this thesis. The CAV technology is applied to develop VSL strategies to improve bottleneck discharge rates and reduce system delays. Three reactive VSL control strategies are developed to enhance traffic stability using: (i) a single CAV (per lane) for control, (ii) a single CAV (per lane) coupled with VMS, and (iii) multiple CAVs. Adaptive schemes to complement the above three strategies are further developed. These VSL strategies using the CAV technology offer significant benefits over conventional control as: (i) they deliver more efficient control by creating a void, which is less restrictive and simpler; (ii) they could be more cost-effective; and (iii) CAVs can serve two key functions simultaneously, traffic monitoring and control action. This thesis formulates probabilistic control failure based on the stochastic traffic instability. This framework is developed based on the probability of instability by individual vehicles in different traffic states to obtain temporal evolution of probability of traffic instability and ensuing control failure. An optimal control speed is determined to maximize the expected delay saving incorporates probabilistic control failures over time.