Strategic Traffic Assignment: Models and Applications to Capture Day-to-Day Flow Volatility

Traffic assignment models continue to play a critical role in the transportation planning process. Furthermore, day-to-day traffic flow volatility is a well-acknowledged phenomenon that planners and researchers alike view as increasingly important. However despite the importance of accounting for volatility, deployed assignment models capable of large-scale application have continued relying on traditional assumptions of determinism and perfect information. This research focuses on the impact of day-to-day demand uncertainty on equilibrium-based traffic models by advancing the concept of strategic traffic assignment. In the strategic user equilibrium (StrUE) model, the daily travel demand is treated as a random variable, and users are assumed to have knowledge about the day-to-day demand but are unaware of the specific traffic conditions they will experience during travel. Therefore, drivers make a strategic route choice to minimize their expected travel cost and follow that route independent of the experienced conditions. The result is an equilibrium assignment based on link flow proportions, as opposed to link flow volumes. Furthermore, as the day-to-day demand realization changes, the equilibrium flow proportions will remain the same. Thus, the resulting flows may appear volatile on a day-to-day basis, but can actually be represented by a higher level mathematical equilibrium.Part I of this thesis explores static models of strategic traffic assignment. Strategic traffic assignment is not only significant as a modelling approach, but also for the implications of the model in important network management applications. Therefore, this thesis implements the strategic traffic assignment model in two common transport problems: road pricing and capacity-enhancement network design. Static equilibrium models are useful for many applications, particularly on a large scale, they cannot capture a number of fundamental traffic characteristics due to their time invariant assumptions. Dynamic traffic assignment is a cutting edge extension to the basic models that provide a more realistic representation of traffic flow, although they are significantly more complex. In order to explore the strategic concept from multiple perspectives, Part II of this thesis proposes the strategic system optimal dynamic traffic assignment (StrSODTA) and explores a network design application.The core contribution of this research is to formulate and explore the implications of the strategic approach to accounting for day-to-day demand uncertainty, and furthermore to demonstrate the impact on practical transport planning applications