Preparation of agent-based modelling input from aggregate 4-step model

Abstract

Agent-based and activity-based modelling are considered as the most suitable means to model the emerging mobility environment (New Mobility Services, Mobility as a Service, ride-hailing, etc.) but the lack of experience on the development of such models as well as their considerable requirements in fine-grained input have hindered their wider adoption. For those reasons, and despite their shortcoming, aggregate 4-step models still constitute the most wide-spread approach in the field of travel demand modelling. This study evaluates the possibility to exploit the outputs of existing 4-step transport models for the preparation of input suitable for agent- and activitybased modelling. In particular, the typical outputs of 4-step models, namely, multi-period OriginDestination matrices (ODs) and traffic assignment results (i.e. link counts) are transformed into home-based trip-chains (i.e. tours). The framework is evaluated on the 4-step model of Larnaka, Cyprus.