University of Twente Student Theses


Improving the Assignment of Bicycle Traffic in a Countrywide Transportation Model by Considering Link Characteristics

Pozdech, M. (2023) Improving the Assignment of Bicycle Traffic in a Countrywide Transportation Model by Considering Link Characteristics.

[img] PDF
Abstract:With cycling being the second most-used mode of personal transport in the Netherlands (Statistics Netherlands, 2022), there is a clear need to be able to predict not only how many cyclists are on the roads right now, but also how their distribution across the network changes when the network changes. This is done today using traffic modeling, with one such model being the OmniTRANS Spectrum (Mobiliteitsspectrum). The project was commissioned by the model’s operator Dat.mobility and they wanted to improve the accuracy of the bicycle traffic assignment in it. While this model can simulate traffic across the whole Netherlands, only the Arnhem-Nijmegen region was investigated in this work. Because the results from this area were to be applied to the whole model, restrictions were put in place such that any adjustments to improve accuracy must be based on attributes that are attached to the links and that the adjustments are done to the link speed values. The investigation performed was done to answer research questions that asked a) if the observed counts of cyclists in this region could be used to determine the impacts of demand factors— characteristics of the link which influence its appeal—on traffic intensities, and b) if the traffic model in question could be made more accurate using the findings from the performed analysis. To achieve this, three different approaches were tried. The first two were unsuccessful and involved regressions attempting to fit the demand factors onto some form of the observed traffic. The final approach—which succeeded in improving the accuracy—involved the use of a Calibration Coefficient (CC). It could identify the features which significantly impacted traffic, as well as scale the speeds to incorporate this impact in the model. This approach involved the use of a calibrated zone-to-zone model to provide observed intensities for near-all links in the node-to-node model being adjusted. Because of this differing construction, a novel metric was developed to validate the results in city centers where significant deviations from reality were observed. It compared the number of cyclists and the distances they cycled within these areas. Ultimately, the adjustments found were not applied to the full model, due to limited time. While the identified CC values, and thus their resultant speeds, could improve the overall distribution of traffic in the full model, the values themselves are not representative of cyclists’ behaviors throughout the whole Netherlands. The CC values are just a representation of the difference between the calibrated and default models within the study area per some characteristic—they give no insight into how appealing a characteristic is for a cyclist.
Item Type:Essay (Bachelor)
Faculty:ET: Engineering Technology
Programme:Civil Engineering BSc (56952)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page