This thesis develops a mathematical framework to solve the problem of dynamic assignment in densely connected public transport (or transit – the two words are interchangeably used) networks where users do not time their arrival at a stop with the lines’ timetable (if any is published).
In the literature there is a fairly broad agreement that, in such transport systems, passengers would not select the single best itinerary available, but would choose a travel strategy, namely a bundle of partially overlapping itineraries diverging at stops along different lines. Then, they would follow a specific path depending on what line arrives first at the stop. From a graph-theory point of view, this route-choice behaviour is modelled as the search for the shortest hyperpath (namely an acyclic sub-graph which includes partially overlapping single paths) to the destination in the hypergraph that describes the transit network.
In this thesis, the hyperpath paradigm is extended to model route choice in a dynamic context, where users might be prevented from boarding the lines of their choice because of capacity constraints. More specifically, if the supplied capacity is insufficient to accommodate the travel demand, it is assumed that passenger congestion leads to the formation of passenger First In, First Out (FIFO) queues at stops and that, in the context of commuting trips, passengers have a good estimate of the expected number of vehicle passages of the same line that they must let go before being able to board.
By embedding the proposed demand model in a fully dynamic assignment model for transit networks, this thesis also fills in the gap currently existing in the realm of strategy-based transit assignment, where – so far – models that employ the FIFO queuing mechanism have proved to be very complex, and a theoretical framework for reproducing the dynamic build-up and dissipation of queues is still missing.