Updated: Feb 14
DUET project partners from Plan4All developed a Proof-of-Concept traffic model for the Flemish city of Bruges. The team obtained road network data from OpenStreetMap to create a traffic model and then deployed it in the TraMod environment to demonstrate the prototype's suitability for operations planning. The results were created as part of the Open Spring INSPIRE Hackathon that ran from March until June 2021.
Challenge 10 of the latest Inspire Hackathon sought to assess OSM’s suitability for traffic modeling purposes. The aim was to find out whether an OSM-derived street network can be used to create a low-cost traffic model which can then be deployed on a server for use by the Traffic Modeller software. To that end, the team followed a five-step process as outlined below.
First, a well defined and topologically correct road network was acquired. The final structure describes the allowed movements within a modelled area of interest, and is meant to make the OSM road network routable, as well as to allow the import of prohibited turns on junctions.
Second, traffic generators representing the supply and demand of car trips in the road network were acquired. Traffic generators have to be connected to one of the network junctions. The number of car trips may come from different demographic sources and should reflect various local parameters such as the number of residents, number of schools and accessibility of public transport. The team took a very rough estimation of the traffic generator’s volume, calculated from the area of a building and the amount of average trips generated by each building. The buildings were then aggregated to the nearest junction and visualised using Voronoi diagrams (see figure). The diagrams were later clustered to reach a significant volume for each generator.
Number of trips per zones in the center of Bruges center (gray color represents buildings and roads, red color - traffic generators areas, centroids and trip numbers)
The next step is to calculate the origin-destination matrix. The OD matrix consists of cells representing the number of car trips from the origin (row) to the destination zone (column). This shows the distribution of the aforementioned supply and domain of traffic generators across the network.
In step four, the OD matrix was assigned to the city street network, while in the last step, traffic was calibrated. The calibration process uses measurements of traffic volumes to confront and adjust the calculated traffic model to the observed reality. Traffic measurements (also called traffic census) are usually acquired at neuralgic points of the traffic network. The data for such a traffic census can be gathered by various sources (e.g. ANPR cameras, built-in road sensors). However, the team was not able to reach a dense enough set of such calibration points. Therefore they used only a general calibration based on the total number of inhabitants in Bruges. For this reason, the absolute numbers of cars displayed in the resulting models can vary from the actual traffic. However the general shape and behavior of the model seems promising and when there is better calibration data available, recalibration should be easy to implement.
The team successfully imported data from OSM into the traffic model and then deployed it as a stand-alone application in the TraMod environment. In the prototype application, road closure scenarios can be experimented with to see changes in traffic compared to the original situation. Simulated traffic conditions can even be explored at the level of particular streets, with red colour representing a potential increase in traffic. Additionally, users can add custom events (e.g. complete closure, partial restriction) and simulate their impact on traffic. Watch the video to see how this works.
If you would like to test the Bruges app yourself, or to discuss how to build a low-cost traffic model for your city, don’t hesitate to get in touch with members of the TraMod team here.