Multi-Agent Traffic Assignment of a Synthetic Stockholm Population

Sammanfattning: Static assignment methods have several limitations which form a motivation for dynamic traffic assignment (DTA). The degree project is aligned with a research project which aims to create a disaggregate model system. The components are the travel demand model Regent, the MATSim simulation framework and a commercial DTA package. The scope of the degree project is the interaction between Regent and MATSim. A synthetic population is generated by Regent. The goal of the degree project is to transform this population onto a form that can be simulated by MATSim and set up a simulation of the greater Stockholm area. A method is developed to assign activity locations to building coordinates and set the parameters of MATSim’s utility function. The model is calibrated using the Cadyts tool and an agent-based dynamic traffic assignment is performed with MATSim. In a case study, road traffic in Stockholm is simulated. Traffic is underestimated but the calibrated model manages to reproduce traffic counts in the morning peak period if there are no incorrect link capacities. The Stockholm bypass is evaluated in an alternative scenario. The result of this evaluation is not reliable due to the prototypical nature of the model system.