Modeling Lane Changes at Freeway On-Ramps With a Novel Car-Following Model Based on Desired Time Headways

Abstract

The traffic flow at freeway on-ramps is influenced not only by the lane changes made by merging vehicles but also by the longitudinal behavior of the merging vehicles and vehicles in the main lane. Existing car-following models are not suitable to represent the longitudinal behavior during merging because they are based on the idea that vehicles intend to reach a steady state, that is, constant time headway and zero speed difference, as soon as possible. At on-ramps, however, merging vehicles have time to reach this steady state until they reach the end of the on-ramp. We therefore derive a novel car-following model based on desired time headways that is able to represent this continuous adaptation toward a steady state. From this car-following model, we derive a lane change model for freeway on-ramps with seven parameters. The lane change model includes a leader selection algorithm, which enables merging vehicles to pass or be passed by vehicles in the main lane. The model also includes components to predict the lane change start time based on surrogate safety measures and to describe the lateral behavior during the lane change. Due to the resemblance to car-following models, the methodology to calibrate the lane change model at the microscopic scale can be adopted from car-following models. To validate the model, we conduct traffic simulations and compare the simulated traffic flow with trajectory data from two German freeway on-ramps. The results show that the model accurately represents the longitudinal driving behavior of merging vehicles and their followers, although it slightly overestimates the number of merging vehicles passing a vehicle in the main lane under congested traffic conditions. The simulations yield accurate headway distributions, except in cases of very risky driver behavior, and realistically capture the macroscopic speed-density relationship at the on-ramp.