VACANCY INTERNSHIP PROJECT Analyzing and implementing the Wright node model Problem description In macroscopic assignment models for road traffic the focus usually lies on link models, that is how traffic propagates on the links in a road network. For models adhering to strict capacity constraints, however, it is of equal importance how to model traffic on the turns of the nodes of the network. Usually, this is done in a first-in-first-out (FIFO) regime and the node model satisfies the requirements formulated in Tampère et al. (2011). However, the FIFO assumption is very restrictive when modelling real-life junctions with inflows that do not influence each other. This yields unrealistic results of traffic assignment models. An alternative node model has been proposed by Wright et al. (2017). Here, the FIFO restriction is dropped and junction inflows and outflows may exhibit varying degrees of independence (Figure 1).
This master project focuses on the mathematical analysis, implementation, and practical performance evaluation of the Wright node model. For the latter, the node model is implemented the Static Traffic Assignment with Queuing (STAQ) assignment model for road traffic. STAQ was developed as an alternative to traditional static traffic assignment (STA) models, providing more accuracy on congested networks without reducing robustness, applicability and accountability and without increasing input requirements, whilst keeping computational requirements to acceptable levels (Brederode et al., 2019). Like STA models, STAQ assumes stationary travel demand during a single study period (typically the AM or PM peak) and instantaneous flow propagation, but contrary to traditional STA models it adheres to strict road capacity constraints.
Master thesis assignment The central question of this project is whether the Wright node model is to be preferred to traditional Tampère-type node models in the STAQ traffic assignment model. The following points will be addressed: 1. Mathematical analysis of both node models, with emphasis on realistic traffic modelling. 2. An implementation of the Wright node model in C++ is provided by the supervisors. This can be used in STAQ, but alternative implementations and variants with attractive properties are welcome. 3. How does the Wright node model change the fixed-point problem of STAQ? And how does it influence the route choice problem around STAQ itself? 4. The implementation of the Wright node model is integrated in a STAQ traffic assignment framework. The following performance indicators should be evaluated compared to a Tampère-type node model implementation: a. (Un)realistic traffic assignment results. b. Convergence speed. c. Computational time.
Research group DAT.mobility Deventer Daily supervisors: Alwin Stegeman, Jeroen van Oorspronk When interested in this Masters thesis assignment, please contact Dr. Alwin Stegeman (astegeman@DAT.nl, +31 (0) 6 15903132)