A mathematical formulation of a dynamic, deterministic traffic assignment algorithm particularly applicable to congested networks is specified. The model lies between the traditional stochastic simulation models, and the static, multicommodity flow formulations of the traffic assignment problem in level of detail. The following properties make the model appear to be useful for the investigation of time varying flows in congested networks of moderately large size. Exogenous demand for travel between trip origins and destinations are treated as piecewise- constant functions of time. These demands are transformed via flow-density relations, which are assumed known for each network link, into piecewise constant functions of distance ("flow packets") which approximate the time and space-varying distributions of vehicle density on network links. The increased densities characteristic of congestion are propagated backward resulting in increased travel times, not only on the under-capacity links, but also on the upstream links feeding them. The model forces the distribution of flows to approach those of "user optimized" flows at any instant by dynamically reassigning flow elements to their shortest-time paths whenever these elements reach nodes which intersect alternative partial paths to their destinations.

  • Supplemental Notes:
    • Sponsored by UMTA.
  • Corporate Authors:

    Stanford University

    Department of Industrial Engineering
    Stanford, CA  United States  94305
  • Authors:
    • Brastow Jr, W C
  • Publication Date: 1973-12

Media Info

  • Pagination: 123 p.

Subject/Index Terms

Filing Info

  • Accession Number: 00098420
  • Record Type: Publication
  • Source Agency: Urban Mass Transportation Administration
  • Report/Paper Numbers: UMTA-CA11-0008-73-10
  • Files: TRIS, USDOT
  • Created Date: Oct 18 1975 12:00AM