Cellular automata approach to three-phase traffic theory

TL;DR

This paper extends cellular automata traffic models to incorporate a two-dimensional steady state region in the flow-density plane, aligning with three-phase traffic theory, and reproduces key congested patterns and phase transitions.

Contribution

It introduces a CA model with a synchronization distance that captures three-phase traffic behavior, improving realism over previous models.

Findings

Reproduces general and synchronized flow patterns

Shows dependence of congestion on flow rates near on-ramps

Analyzes capacity drops due to congested pattern formation

Abstract

The cellular automata (CA) approach to traffic modeling is extended to allow for spatially homogeneous steady state solutions that cover a two dimensional region in the flow-density plane. Hence these models fulfill a basic postulate of a three-phase traffic theory proposed by Kerner. This is achieved by a synchronization distance, within which a vehicle always tries to adjust its speed to the one of the vehicle in front. In the CA models presented, the modelling of the free and safe speeds, the slow-to-start rules as well as some contributions to noise are based on the ideas of the Nagel-Schreckenberg type modelling. It is shown that the proposed CA models can be very transparent and still reproduce the two main types of congested patterns (the general pattern and the synchronized flow pattern) as well as their dependence on the flows near an on-ramp, in qualitative agreement with the recently developed continuum version of the three-phase traffic theory [B. S. Kerner and S. L. Klenov. 2002. J. Phys. A: Math. Gen. 35, L31]. These features are qualitatively different than in previously considered CA traffic models. The probability of the breakdown phenomenon (i.e., of the phase transition from free flow to synchronized flow) as function of the flow rate to the on-ramp and of the flow rate on the road upstream of the on-ramp is investigated. The capacity drops at the on-ramp which occur due to the formation of different congested patterns are calculated.