Network-based models have been central to planning and managing urban transport systems for more than four decades. In particular, in road transport, assignment and simulation models are the primary tools used for understanding the routes drivers choose and the impacts they have on traffic levels and travel conditions in time and space.
However, modelled representations of urban road networks have traditionally been incomplete and selective, focussing on just those links and junctions which are considered to be of interest to policymakers and which are compatible with available zonal aggregations of travel demand. Over time and in parallel with increased computing power for handling data and running models, the spatial scales of modelled networks and the proportion of the real network represented have tended to increase. This has had the implication of increasing the complexity of modelled networks, making them potentially both harder to interpret and less flexible for scenario testing. Even so, the network representation is still not comprehensive and the process through which links and junctions are included or excluded is typically based on ad hoc user choices rather than any analytical methodology.
In addition, the emergence of new data sources such as vehicle tracking from mobile phone records and Bluetooth devices presents both challenges and opportunities for transport network modelling. On the one hand, the data produced naturally provides greater spatial detail than has been available previously including comprehensive coverage of networks, perhaps suggesting that modelled representations should also progress in that direction. On the other hand, the continuous availability of a data stream over time has the potential to allow a much wider range of scenarios to be observed and understood, suggesting that models may also be used and refined in a more continuous manner to provide the facility to develop more sophisticated understandings and predictive capabilities. These possibilities may be seen as pulling in opposite directions and lead to the critical questions of just how much spatial and network-based detail is required to provide useful insights for planning and policy and how, for any given case study, the selection of appropriate levels of detail and aggregation may be made without simply relying on user perceptions.
Only a small amount of work has been carried out relevant to this topic previously, so it seems well positioned to add to knowledge and to produce novel and useful outcomes. There is a long tradition in transport network modelling research of using simplified network structures to investigate core principles and new ideas, but these networks are typically synthetic with few similarities to real cities. Milne (1998) used an experimental methodology to develop a simplified version of a larger modelled network for sensitivity testing purposes during a study to investigate the network effects of urban road pricing systems. This succeeded in demonstrating the potential of a simplified network structure for replicating the trends observed in the larger network on which it was based. But it did not have access to significant volumes of observed data and was therefore unable to investigate the relationship between the modelled networks and reality. It also lacked a formal analytical methodology. More recently, Anastasiadis (2015) has reviewed a range of methodologies for creating aggregated network representations and attempted a preliminary application of one approach. Again, the main focus was on developing a simplified version of a larger modelled network, but in this case with access to a significant volume of observed traffic flow data for two distinct scenarios. The initial results were mixed, but suggested considerable promise for the development of more sophisticated methods during a longer study. Separately, Frejinger and Bierlaire (2007) have developed a subnetwork approach for attempting to solve some of the difficulties encountered in applying random utility models to route choice problems. While the context is rather different, this work may be useful in widening the scope of approaches available for analysing the implications of incomplete network representation. In addition, von Landesberger et al (2015) have addressed the issue from an alternative direction, using mobility data for Twitter users in Greater London as a means of defining spatial movement networks at different levels of detail. Developing an understanding of the relationship between these approaches and the observed data and associated models pertaining to urban transport networks should make a valuable contribution to furthering understanding of how modelling for transport planning and policy purposes needs to evolve to make best use of new data sources.
Please visit our LARS scholarship page for more information and further opportunities: https://www.environment.leeds.ac.uk/study/postgraduate-research-degrees/lars-scholarships/
The project would be particularly suitable for someone from a numerate background who has studied the basics of transport network modelling at Bachelor or Masters level. Having gained some experience of the use of transport network models in practice would also be an advantage.
Anastasiadis, P. (2015). Development of an aggregated network and testing its efficiency in reflecting changes in traffic patterns. Masters thesis, Institute for Transport Studies, University of Leeds, unpublished.
Frejinger, E. and Blerlaire, M. (2007). Capturing correlation with subnetworks in route choice models. Transportation Research Part B, 41(3) 363-378.
Milne, D.S. (1998). Modelling the network effects of urban road-user charging. PhD thesis, Institute for Transport Studies, University of Leeds, unpublished.
von Landesberger, T., Brodkorb, F., Roskosch, P., Andrienko, N., Andrienko, G. and Kerren, A. (2016). Mobility Graphs: Visual Analysis of Mass Mobility Dynamics via Spatio-Temporal Graphs and Clustering. IEEE Transactions on Visualization and Computer Graphics 22(1) 11-20.