Modelling and Microsimulation of Activity Generation, Activity Scheduling and Mobility Assignment

Abstract

This thesis develops a novel framework for modelling and microsimulation of activity-based travel demand. It explores alternative micro-behavioural modelling methods for individuals’ activity participation, time allocation, shared travel choice, mode choice and vehicle allocation. For example, mixed logit models of shared travel choices are developed that accommodate individuals’ social interactions with household and non-household members while travelling to different activity-based tours. A multiple discrete continuous extreme value model is formulated, which addresses individuals’ social interactions within the modelling framework to explore the joint decision of activity participation and time allocation. Latent segmentation-based random parameter logit models are developed to evaluate vehicle allocation decisions for different activity-based tours. This thesis also presents the development of a novel activity-based shorter-term decisions simulator (SDS) to predict activity and travel decisions. SDS consists of three sub-modules: activity generation, activity scheduling and mobility assignment, which are developed by implementing different components, namely activity types, frequencies, durations, start times, destination locations, shared travel arrangements, mode choice and vehicle allocation. The model addresses underlying process mechanisms of such components within the microsimulation framework by developing advanced modelling techniques. For instance, activity generation implements a Markov Chain Monte Carlo method to represent the process orientation of generating activity types. Activity scheduling is implemented as a three-stage decision process: activity agenda formation, destination location choice and shared travel choice, and accommodates social interaction-based feedback from shared travel choice component within the computational procedure. Furthermore, mobility assignment is a two-stage dynamic process of mode choice and vehicle allocation that addresses social interactions within empirical and computational procedures. SDS generates baseline information for the year 2006, and simulates activity and travel decisions for a 30-year period of 2007-2036. This thesis presents the validation of the SDS model results, and predicts the evolution of activity-travel information of Halifax population. SDS is implemented within the integrated urban model, iTLE, and provides activity-based feedbacks to households’ long-term residential decisions; thus, develops an integrated and behaviourally consistent urban modelling system. The SDS microsimulation model would be helpful to test different emerging travel demand management strategies as well as alternative land use and transportation policy interventions.