Empirical Assessment of a Markovian Traffic Flow Model

Abstract

Previous research efforts on developing traffic flow models to account for traffic flow dynamics over transportation networks have centered on macroscopic high-order models. It is unclear whether traffic flow dynamics can be well described using a high-order model formulation, but for real-time traffic control, it is important to have tractable yet accurate models. Described here is a set of tractable traffic flow models based on the Markovian compartment concept. The basic models can be further modified to produce effects analogous to high-order models in capturing unstable traffic behavior during congestion. Special treatments are also made to account for the effects of conflicting flow on the predicted turning exit flow at intersection approaches. The proposed models are evaluated using field data. The results indicate that all the model parameters, including traffic flow parameters and gap acceptance parameters, are reasonably estimated, and the underlying models provide good fits to the field data.