Cellular Automata Model for Work Zone Traffic

Abstract

This paper proposes a cellular automata (CA) model incorporating work zone configuration to model work zone traffic. The randomization probability parameter of the proposed CA model is able to characterize driver acceleration–deceleration behavior. The randomization probability should be a function of traffic flow and work zone configuration that comprises the activity length and transition length; however, past studies have assigned a hypothetical constant value for the randomization probability. This paper calibrates the randomization probability from field data, which is determined by minimizing the square error between simulated travel time and observed travel time by using a trial-and-error method. A polynomial regression method is employed to formulate the randomization probability functions in and outside the work zone. A case study was performed to test the proposed CA model dependent on work zone configuration. Comparison of field data and the proposed CA model for travel time and traffic delay shows very close agreement. Statistical comparison of the simulated results from the proposed CA model and PARAMICS indicates that the proposed CA model performs well in modeling work zone traffic.