A utility-based dynamic demand estimation model that explicitly accounts for activity scheduling and duration. (August 2018)

• Record Type: Journal Article
• Authors: Cantelmo, Guido; Viti, Francesco; Cipriani, Ernesto; Nigro, Marialisa
• Is Part Of: Transportation research. Volume 114(2018)Part B
• Page Start: 303
• Page End: 320
• Publication Date: 2018-08
• Subjects: Offline Dynamic Demand Estimation -- Dynamic Traffic Assignment (DTA) -- Utility functions -- Departure Time Choice model -- GLS -- SPSA; Transportation -- Research -- Periodicals; 388.011
• DOI: 10.1016/j.tra.2018.01.039
• Languages: English
• ISSNs: 0965-8564

Abstract

Highlights: The proposed formulation systematically reduces the number of decision variables. A richer information can be contained in the generated demand matrices. The localism of the problem is reduced, hence the reliability increases. The proposed formulation is more sensitive to the spatial distribution of the demand. Activity duration increases demand observability over both time and space. Model properties are tested with a realistic trip-based macroscopic DTA model. Abstract: This paper proposes a Dynamic Demand Estimation (DODE) framework that explicitly accounts for activity scheduling and duration. By assuming a Utility-Based departure time choice model, the time-dependent OD flow becomes a function, whose parameters are those of the utility function(s) within the departure time choice model. In this way, the DODE is solved using a parametric approach, which, on one hand, has less variables to calibrate with respect to the classical bi-level formulation while, on the other hand, it accounts for different trip purposes. Properties of the model are analytically and numerically discussed, showing that the model is more suited for estimating the systematic component of the demand with respect to the standard GLS formulation. Differently from similar approaches in literature, which rely on agent-based microsimulators and require expensive survey data, the proposed framework is applicable with all those DTA models, which are based on OD matrix, and do not necessarily need any data at user level. This has been proven by applying the proposed approach with a standard macroscopic realistic Dynamic Traffic Assignment (DTA).