MULTIPATH TRAFFIC ASSIGNMENT: DEVELOPMENT AND TESTS

The assignment technique described consists of a stochastic process in which individual trees are found according to minimum time paths, using link times which vary from the specified link time by randomly selected amounts. (A tree is the set of minimum time paths from one origin to all possible destinations.) A different set of link times is used for each tree built thus enabling several different paths to be found between the same points. The method was used for tests on unrestrained assignments to three networks with differing characteristics. Comparisons were made with observed data and conventional minimum path assignments for those networks. In all three cases the multipath technique was found to produce link volumes significantly closer to the observed values than did the minimum path assignment. The computer program also permits the use of several methods of capacity restraint which may be used either with the multipath assignment or with conventional mimimum path assignment. The alternative methods are described together with the results of a series of tests to determine their effectiveness when used with each assignment technique. The capacity restraint techniques may be divided into two categories which are defined as interative and incremental. With both techniques better results were obtained in conjunction with the multiple route assignment than with the minimum path. In the case of the iterative technique, it was found to be essential to use the multipath assignment in order to get good results. In comparing assigned and observed link volumes there was little difference found between the methods of capacity restraint, but the iterative method, used in conjunction with the multipath assignment, was found to produce more realistic link speeds and travel time estimates. The results of this method, however, were more sensitive to the selection of the correct values of parameters than were the incremental methods. The multipath assignment uses more computer time than does a mimimum path assignment but the resulting difference in cost is not thought to be significant in a transportation planning study.

  • Corporate Authors:

    Ontario Ministry of Transportation & Communic, Can

    Research and Development Division, 1201 Wilson Avenue
    Downsview, ONo M3M 1J8,   Canada 
  • Authors:
    • Dalton, P M
    • Harmelink, M D
  • Publication Date: 1974

Subject/Index Terms

Filing Info

  • Accession Number: 00125467
  • Record Type: Publication
  • Report/Paper Numbers: RR189
  • Files: TRIS
  • Created Date: Dec 3 1975 12:00AM