Effectiveness of non-directional lane strategy at signalized intersections: A cellular automata study

In the future urban road network, regular vehicles (RVs) and connected vehicles (CVs) will coexist for the long term. Lane direction assignment and lane change strategies in intelligent and connected environments will significantly impact intersection efficiency and safety. This study focuses on a double-lane signalized intersection approach and develops a microscopic simulation framework for mixed traffic flow using a Cellular Automata (CA) model. We propose a conflict-based lane change strategy with no lane direction assignment (CV-CLC) and direction-based lane change strategies with lane direction assignment (CV-DLC and CVRV-DLC). The model refines the acceleration and deceleration rules, random deceleration probability, and lane change safety conditions. Numerical experiments compare traffic characteristics under these strategies, focusing on time-space diagrams, flow arrival rate relationships, speed arrival rate relationships, delay time, average lane changes per vehicle, and the number of vehicles failing to complete a turn. The results validate the effectiveness of the CV-CLC strategy. Furthermore, the findings indicate that fixed lane direction assignment strategies are insufficient for handling real-time traffic flow variations, while the CV-CLC strategy is better equipped to adapt to dynamic traffic conditions. Among the three strategies, the CV-CLC strategy results in the smallest average vehicle delays. Moreover, the CVRV-DLC strategy results in the fewest number of turn failures, which suggests that allowing only CVs to change lanes has certain limitations. However, an increase in CV penetration helps mitigate this issue. This study provides theoretical insights for lane design and control optimization in the era of intelligent and connected vehicles.