Flexible lane allocation for dedicated automated vehicle zones

As urban traffic demands generally fluctuate a lot during different periods of a day, current fixed lane allocation may not be able to provide satisfactory service for the varying and imbalanced traffic flows. With the emerging connected and automated (CAV) technologies, this study proposes a joint flexible lane allocation model to improve the network service capability. Different from previous studies, the model optimizes not only the lane reversals among adjacent intersections but also the lane groups in front of intersections. Moreover, the bottleneck effect of intersections has been thoroughly considered in the modelling process, preventing the system from falling into local oversaturation status. To achieve solutions in reasonable time, a bi-level framework enhanced by linearization techniques is proposed to decompose the coupling relationship between lane allocation and traffic assignment and resolve the curve of dimensionality for large scale networks. Under the framework, solutions can be obtained by iteratively solving two mixed-integer linear programming (MILP) models, determining the lane allocation results and traffic flow assignment respectively. Numerical analysis on an artificial network is provided to illustrate the effectiveness of our proposed method; furthermore, numerical experiments are conducted on a real-world network and validate that our proposed method can outperform state-of-practice lane allocation methods.