EQUILIBRIUM TRIP ASSIGNMENT: ADVANTAGES AND IMPLICATIONS FOR PRACTICE

During the past 10 years the problem of assignment of vehicles to large, congested urban transportation networks according to the principle of equal travel times has been solved and an efficient, convergent computer algorithm devised. Although the algorithm is available in the Urban Transportation Planning System, many practitioners continue to use the heuristic trip-assignment algorithms devised in the early 1960s. As in many other cases, this slow implementation of a new, improved algorithm appears to come from (a) a lack of understanding of its basic concepts, (b) an unfamiliarity with the computer program for applying the algorithm, and (c) a lack of evidence concerning the new algorithm's performance in large-scale applications. These three issues are addressed in this paper. Based on the experience with its implementation on a large network, it is recommended that equilibrium trip assignment should always be used instead of iterative assignment. Better results, as judged by the criterion of equalizing travel times for alternative paths between each origin-destination pair, will always be obtained with the equilibrium algorithm for any given amount of computational effort. Which method best replicates the observed vehicle flows may depend on the detail of the network, the adequacy of the capacity-restraint functions, and the time period of the assignment (24 h or peak period). (Author)