The paper discusses a multiple route model in which it is assumed that the person making a trip in a network does not know the actual journey time on each link of the network, but that he associates with each link a supposed time, which is drawn at random from a distribution of times having as its mean the actual link time. The person then finds and uses the route which minimizes the sum of his supposed link times. This proves to be a powerful method of obtaining realistic multiple routes between pairs of nodes. The procedure can be used for making assignments in unrestrained situations, or where capacity restraint is required, it is found that convergence of iterations is obtained much more quickly using multiple routes, than using single routes.
Samenvatting