One of the major challenges in the traffic assignment field is to properly take into account capacity constraint problems. This will allow the analysts, in their simulation procedures, at least to avoid traffic propagation through bottlenecks, to most properly evaluate the travel times between the O/D couple of study area and to adequately compute the performances of the traffic network. Moreover an assignment procedure able to take into account capacity constraints allows designing and simulating traffic schemes oriented to "protect" city centres from excessive traffic flows. Capacity reduction strategies (applied to properly designed "centre-surrounding" cordons), in fact, can be adequately evaluated. The analysts have made most of the modelling efforts related to capacity-constrained assignment on Dynamic Network Loading procedures or on static, but deterministic, assignment models. Provided that the behavioural assumptions of the Deterministic Assignment models are proved to be often unsatisfactorily and that Dynamic Loading Models are still out of easy and current practical application in the proposed paper a static but behaviourally consistent assignment model is presented. The proposed model is analytically formulated in rigorous terms, the study of enhanced cost functions is made and a suitable assignment algorithm is presented. The capacity-constrained stochastic assignment model is applied to a test-network of small size in order to easily present the model capabilities. The simulation of the capacity-constrained stochastic assignment is also made on a real size network in order to test the model's applicability.
Abstract