Abstract Manager

Activity-based travel demand model systems (ABMs) are increasingly common in Canada, with operational or near-operational ABMs in the Greater Toronto-Hamilton Area (multiple implementations), Montreal, Vancouver, Ottawa and Halifax, advanced trip/tour-based model systems in cities such as Calgary and Edmonton, and active ABM research programs in a number of Canadian universities, such as Dalhousie, Polytechnique Montréal, UBC and University of Toronto, among others. These model systems include a number of common features, including:

• A common overall system structure key elements of which are: population synthesis, medium/long-term choices (place of work, auto ownership, etc.), a daily activity/travel model, and a network model for road and transit assignments.
• They are developed based on similar household travel surveys, augmented by similar count data (screenline counts, transit station boardings, etc.).
• Use of a commercial network software platform for road and transit assignment, often Emme (now known as OpenPath), although other platforms, such as Visum and Aimsun, are also used, or less commonly, open source platforms such as MATSim.

Despite these commonalities, each urban area’s model system is still largely bespoke, requiring extensive system development and calibration efforts for each implementation. There is also a fair bit of “reinventing of wheels” as the same lessons tend to be learned and relearned in each implementation. And, small/medium-sized urban areas are generally unable to implement ABMs due to lack of data, cost of commercial platforms and/or (often) technical expertise.

The question of model transferability is as old as the discrete choice modelling methodology that underlies all current models, with results that are often encouraging but not yet broadly conclusive. If a generally transferable model system (requiring, at most, relatively “lightweight” local calibration), based on commonly available data (StatsCan and otherwise) and a common, flexible software platform were available in Canada, this could dramatically improved Canadian planning practice and support more comprehensive and consistent policy analysis and decision-making.

This paper explores the proposition of developing a transferable activity-based travel demand model system for use by Canadian urban regions. It discusses the data, modelling methods and software available to support such a transferable model system, as well as major challenges to this work, and it sketches a possible work plan for testing the transferability hypothesis.