Computational Modelling Group

Seminar  19th December 2011 4:15 p.m.  Building 22, 1st Floor Lecture Room, Highfield Campus

Formulation of a risk-averse system optimal dynamic traffic assignment model

Prof Mike Bell
Imperial College

Categories
Transport
Submitter
Elizabeth Roe

Prof Mike Bell, Imperial College London

Dynamic system-optimal assignment is recognized in literature as an important area of research, with potential applications in the areas of traffic control, road pricing and policy. System-optimal assignment can provide valuable insights for traffic managers as it indicates how transport systems can be improved.

This paper describes a system-optimal dynamic traffic assignment problem that incorporates the concept of strategies in traffic assignment as a way of managing risk. The inherent uncertainty within traffic networks is mitigated through the adoption of a risk-averse approach in the assignment of road users. In the implementation presented in this paper, assignment is directed by a risk-averse network controller and seeks to minimize the pessimistically expected travel time by minimizing total travel time plus the exposure to exceptional delay (due to unforeseen events like adverse weather conditions). This involves the provision of alternative routes should events occur in the network that result in some links becoming unavailable or unattractive.

A three-pass algorithm is formulated which calculates the hyperpath set and performs a dynamic network loading. The efficacy of the algorithm is demonstrated by a series of propositions and through computational experiments. The implications of the topology, size and other features of road networks are also discussed. The environmental benefits of applying the strategy described in this paper are also addressed, in terms of congestion reduction.

All are welcome.

Prof. R. Eddie Wilson
Transportation Research Group
University of Southampton
Telephone: (023) 8059 2174
Facsimile: (023) 8059 3152
Email: R.E.Wilson@soton.ac.uk