Erik Roos presenterar sitt examensarbete vid Rymd-, geo- och miljövetenskap "Traffic Congestion in Urban Modelling".
Handledare: Alexander Hellervik, Trafikverket.
Examinator: Claes Andersson, Fysisk resursteori, Chalmers.
Opponenter: Maximilian Ludvigsson, Oscar Kalldal.
In this thesis a model for traﬃc congestion in urban areas is developed using betweenness centrality. The betweenness centrality is augmented with the activity in the network, which corresponds to the assessed value of a property. In this way shortest paths between two nodes with high activity contribute more to the overall betwenness centrality, and in turn traﬃc, than between two nodes with low activity. The traﬃc network is imported from Open Street Map.
In each iteration in the model the betwenness centrality in the network is calculated and the speeds on the edges are updated according to the VQrelations. A new activity is then calculated with the new speeds. Travel times in the network are ﬁnally compared to travel times given by Google.
The model is tried out for a number of settings in the city of Gothenburg. The travel times obtained with the speed limits given from OSM show high linearity but the average travel time is too short. The model manages to improve the travel times to some extent but looses linearity when too much traﬃc is introduced. Updating the activity in each iteration shows little diﬀerence to calculating the activity once and keeping it static. When comparing ﬂows, the model does indeed capture some real life behaviours but fails to capture others. For example in the inner city or along the major highway going through Gothenburg the traﬃc is not slowed down at all in the model.
Future work includes studying diﬀerent ways of deﬁning the shortest path between nodes (such as topological or angular distance), introducing some through traﬃc or increasing the geographical calculation area and increasing the resolution of the traﬃc network (making it directed and have more precise road types).
EL42, lecture room, Linsen, EDIT