Behavioural Modelling of Urban Freight Transportation: Activity and Inter-Arrival Duration Models Estimated Using GPS Data

This dissertation details the development of two behavioural models of urban commercial (freight and services) transportation: activity duration and inter-arrival duration models. These models were designed to comprise two components within the proposed FRELODE (FREight LOgistics DEcisions) modelling framework; FRELODE accepts a list of shipments or service trips by carrier and models how carriers execute urban shipments, creating a series of resulting vehicle movements (trips and tours). My Ph.D. research produced significant progress in four areas. First, the activity and inter-arrival duration models were estimated using a three-month passively-collected GPS dataset from truck-mounted GPS-equipped engine-on-board recorders. To the author's knowledge, this research represents the first attempt in the literature to estimate components of a commercial-vehicle travel demand model using passively-collected GPS data as the primary data source. Exploring new data sources, such as GPS, is important because little data describing urban commercial vehicle movements are currently available. New data processing techniques were developed to convert the longitudinal GPS data into a travel diary suitable for transportation model estimation. Second, two hazard models of commercial activity duration were estimated. Third, models of inter-arrival duration were estimated using the longitudinal GPS dataset to model the number of days between repeated visits to the same destination. To the author's knowledge, these represent the first estimated models of inter-arrival duration for commercial transportation. The fourth area of significant process described in this dissertation is the development of the FRELODE modelling framework. FRELODE considers a multiple day time period as commercial establishments do not necessarily operate using consistent schedules, and also accounts for carriers delivering multiple shipments in a single tour. In conclusion, passively-collected GPS data were found to hold promise as a complement to existing data sources of commercial vehicle travel. Also, it is expected that these estimated models, included within the FRELODE framework, will form components within a larger proposed agent-based microsimulation commercial vehicle modelling framework that is currently under development.