Trip lengths and the macroscopic traffic simulation: an interface between the microscopic and macroscopic networks
Résumé
Macroscopic traffic simulation has gained more and more interest from the scientific community since the seminal works of Daganzo (2007) and Geroliminis & Daganzo (2008). It assumes that traffic conditions are homogeneous inside a given region (or reservoir); and traffic states are defined by a Macroscopic Fundamental Diagram (MFD). Most implementations of MFD models are accumulation-based and have been designed for control purposes. Few attention has been paid to the route choice process, i.e. the characterization of most likely alternatives and calculation of pathflow distribution. The most advanced MFD-based simulator including route choices has been proposed by Yildirimoglu & Geroliminis (2014). The authors implemented a multi-reservoir accumulation-based macroscopic traffic simulator. To calculate the macroscopic routes and trip lengths, they considered the sampling of several microscopic origin-destinations (OD) on the real network and made use of a time dependent shortest-path algorithm. The route flows are assigned according to a simple Multinomial Logit model. In fact, Multinomial Probit model has never been implemented in such a framework, while lots of route correlations happen due to the limited number of reservoirs used to describe an urban area. Furthermore, the macroscopic route set definition still requires further investigations.
Origine : Fichiers produits par l'(les) auteur(s)