Stability and Performance of Car-Following Models in Congested Traffic

Abstract

Car-following models NETSIM, INTRAS, FRESIM, and CARSIM have been extensively used for simulation of traffic conditions, including congested conditions. The car-following algorithms of these models are well known and available in the literature. However, the stability, performance, and characteristic car-following behavior of these models have not been investigated in the literature. Since the model outcomes are used in place of real data in most circumstances, it is important to know the individual vehicle behavior and car-following stability of these models. In this paper, the car-following algorithms of the above mentioned models and a recently introduced model, INTELSIM, are examined for their stability and performance in congested traffic conditions. The performance of the car-following models and the individual vehicle behavior are evaluated for accuracy, stability, problem areas, and overall model behavior for the congested conditions. Moreover, this paper aims to help the users utilize the models correctly and understand the outputs better by showing insights to the individual vehicle behavior simulated in the above models. It was found that vehicles in NETSIM and CARSIM models car-follow at separation time headways approximately equal to their reaction times. In addition, INTRAS and FRESIM presented unrealistic acceleration fluctuations and a large number of maximum decelerations. INTELSIM model presented speed and spacing profiles similar to those of the drivers.