| Abstract: |
Traffic congestion at freeway on-ramps and off-ramps presents a major challenge in urban transportation planning, especially during peak hours and emergencies. Managing traffic flow and understanding driver behavior at these points is critical for improving efficiency and safety. Various methodologies have been used to coordinate freeway segments and connecting roads. However, most studies focus on control strategies like ramp metering and merging techniques to manage vehicles entering the freeway, while off-ramp traffic control has been largely overlooked. Queue formation at exit ramps can disrupt mainline traffic and extend congestion upstream, particularly during peak hours. Effective traffic management at exit ramps ensures smooth transitions to local roads, prevents congestion, and improves safety at intersections. Unlike on-ramps, ramp metering is not typically used for off-ramps. Instead, strategies such as traffic signal control, queue management, and interchange design adjustments help regulate exiting traffic. The focus for off-ramps is on ensuring smooth and efficient exits and merging rather than controlling freeway entry. Traffic signal control at off-ramps manages vehicle flow at intersections where ramps meet local roads. While effective, it requires precise timing and struggles with fluctuating conditions. Traditional rule-based approaches, like speed limits and lane designations, rely on predefined rules but lack adaptability. Ad hoc approaches, such as temporary diversions, provide situational solutions but are not long-term fixes. More recently, machine learning-based reinforcement learning (RL) methods have gained popularity for real-time traffic adaptation. However, RL models require periodic retraining, increasing computational demands and predictability challenges. In contrast, analytical methods like queuing theory are well-suited for managing off-ramp traffic in diverse road networks. Queuing models capture human driver behavior, crucial for optimizing exit ramp flow, yet remain underexplored. Queuing theory excels in predictive modeling, optimizing traffic flow through systematic vehicle pattern analysis. Its computational efficiency makes it a practical, scalable solution for dynamic traffic environments. Traditional queuing models, however, face limitations. They struggle with complex behaviors like balking, where drivers avoid long queues, and reneging, where drivers leave the queue after joining. Standard models assume traffic demand does not exceed capacity (ρ < 1), making real-world congestion difficult to model. We propose leveraging queuing theory to balance adaptability and efficiency in off-ramp traffic management. Specifically, our model will analyze balking and reneging while also investigating reneging retention, where drivers reconsider and rejoin the queue. These behaviors will be incorporated into queuing equations and simulated under varying conditions to assess their impact on overall traffic flow. This enhanced queuing model offers a robust framework for performance evaluation, improving traffic efficiency and safety. |