An Agent-Based Simulation Model for Optimization of the Signalized Intersection Connected to Freeway On-Ramp

TL;DR

This study develops an agent-based simulation model to optimize signal timing at a complex intersection connected to freeway on-ramps, proposing and testing solutions to reduce delays and improve safety.

Contribution

It introduces a VISSIM-based agent simulation for intersection optimization and evaluates three novel lane management strategies.

Findings

Solution (3) significantly reduces vehicle delay.

Adding a diverge lane improves traffic flow.

Simulation results favor lane reconfiguration over lane lengthening.

Abstract

Unlike most existing studies on off-ramp traffic signal control, this paper focuses on the optimization problem of the signalized intersection connected to freeway on-ramps. Conflicts are often observed between the traffic heading to an on-ramp and the traffic continuing straight which leads to issues such as intersection overflow, increased delay, and concerns about safety. For studying this problem, a real-world signalized intersection in Buffalo, New York was chosen, which has two through lanes and one short shared (through and right-turn) lane. At the downstream of the intersection are two following on-ramps, one to the highway I-290 West and the other to I-290 East. During peak hours, the shared lane often observes a long queue, which furthermore blocks the through traffic on the parallel lane. To solve this problem, a VISSIM agent-based simulation model was built and calibrated based on field observations. Three potential optimization solutions were proposed and tested with the help of VISSIM: (1) increasing the length of the short shared through and right-turn lane; (2) making the short shared through and right-turn lane right-turn only, and (3) adding a new diverge lane for the right-turn vehicles. According to the simulation results, solution (3) performs the best, resulting in the least vehicle delay time.