Dynamic user equilibria in StreamLine

Dijkhuis, N.H. (2012) Dynamic user equilibria in StreamLine.

Abstract:StreamLine is a dynamic traffic assignment framework in OmniTRANS, a traffic assignment model that is developed by Omnitrans International. StreamLine tries to determine a dynamic user equilibrium (DUE). In this research a DUE is defined as follows: ‘An equilibrium arises when for each route choice moment, for each OD pair, the flow unit costs on the utilised routes are equal to the minimum unit path cost at the corresponding point in time’. The goal of the research is to investigate to which extent StreamLine is able to determine a DUE and to look to which extent this equilibrium changes due to parameter changes in StreamLine. This is done by designing test networks in StreamLine, running some simulations with different conditions, and analysing the results of these simulations. Whether a DUE can be determined, depends on the method how StreamLine determines route fractions during the simulation. When a PCL assignment is used, it is not possible to determine a deterministic equilibrium, since always a small part of the demand is assigned to each route. In theory, this is possible with a series of all-or-nothing assignments. However, the usage of an all-or-nothing assignment elicits some problems in MaDAM, the propagation model in StreamLine, that makes it impossible (for the designed networks in the research) to reach a realistic DUE. Unrealistic situations arise during the emptying of the network, when large density differences arise near merge nodes, and when the flow (or demand) on a link changes abruptly. These problems can probably be prevented by doing some adaptations to the propagation model. Hereafter, some parameters are varied. The length of the route choice interval influences the accuracy of the results. The shorter a route choice interval, the more realistic the development of the route costs will be, but there are more iterations needed before the route costs are converged. In the parameter analysis, it became clear that the parameters tau, kappa and nue do influence the equilibrium, but the changes in route costs are quite small. Kappa and nue influence the equilibrium more than tau. It is possible that the effects are different when a larger and more complex network is used. During the analysis, unrealistic results occurred due to the oscillation effect, which means that the route fraction distribution during the simulation time is not stable, and the exceeding of the jam density on some links. It is recommended to make some changes to the propagation model. To prevent the observed problems, the anticipation term in MaDAM must be changed, StreamLine must calculate route costs based on average speeds instead of densities, and it must not be possible that the density that is calculated by MaDAM exceeds the jam density.
Item Type:Essay (Master)
Omnitrans International
Faculty:ET: Engineering Technology
Subject:55 traffic technology, transport technology
Programme:Civil Engineering and Management MSc (60026)
Link to this item:http://purl.utwente.nl/essays/62157
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page