The paper analyzes the steady-state behavior of car-following models within state-of-the-practice commercial traffic simulation software. The car-following models are classified based on their uncongested regime steady-state behavior into two categories. Apart from the INTEGRATION software that employs the Van Aerde car-following model, the research demonstrates that all state-of-the-practice traffic simulation software assume a constant desired speed that is insensitive to the flow level within the uncongested regime. The paper then quantifies the effect of speed variability, as a common approach to model driver differences with respect to desired speed, on the steady-state traffic stream behavior. The paper not only demonstrates that the speed variability has a significant impact on the speed-at-capacity, but also develops procedures for the calibration of the steady-state relationship while accounting for speed the effects of speed variability. Finally, the effectiveness and validity of the proposed procedure is demonstrated through an example illustration.
Abstract