Predicting Travel Patterns Under Disruption and Change

Project reference: LE76

Application deadline: 4th July 2023

Click here for instructions on how to apply

*This opportunity is only open to Home rated applicants (UK citizens or those with indefinite leave to remain in the UK)*

Disruptions, events, incidents, and closures are daily characteristics of urban transport systems. Yet, our understanding of how people change their travel behaviours in each situation is poorly understood. This is partly due to the ad hoc nature of these events – some closures being planned and advertised well in advance, while others being completely unexpected incidents – as well as the uncertainty surrounding the impact of closures, reliability of information, and the best available alternative. Improving our understanding of behaviour under different scenarios, and development of a robust model of behaviour, would be beneficial to transport authorities seeking to promote behaviour change to improve services.

This project will undertake analyses of changes in travel behaviour following disruption and develop new agent-based models that predict the outcomes of future disruptions. The project will co-supervised by colleagues at Transport for West Midlands (TfWM). The aim of the collaboration will be to provide real-world application and context for the modelling, in addition to promoting knowledge exchange. It is anticipated the researcher will spend some time spent in placement with the Data Insight Service Team. It is envisaged there will be three main stages to the project.

Stage 1 – Behavioural analysis under disruption

In this first stage, we will review prior research literature to update and extend taxonomies of disruption (e.g. Zhu and Levinson 2010, Marsden et al., 2020), and behaviour change, considering the role of prior information (e.g., messaging campaign, media reports), travel mode, location, time of event, and other relevant factors, on changes in travel mode choice (e.g. Schaefer et al., 2021), route choice (e.g. Marra and Corman, 2023), and departure time (e.g. van Exel, et al. 2009, Rahimi et al, 2019) . With this framework in place, we will use mobility data sources to analyse aggregate changes in demand under disruption incidents and assess the extent to which mitigating interventions had on adjusting demand. We will also consider how the impacts of service disruptions persist through time, establishing which lend themselves to longer-term (‘sticky’) adjustments to behaviour. This initial analysis will undercover new information on how responses vary in reaction to different types of events, it will also inform the subsequent modelling stages.

Stage 2 – Modelling behaviour under disruption

Informed by analysis of prior disruption events, we will then consider how to model and predict behaviours future under disruption. This stage of work will build on the latest computational modelling approaches and qualitative theories of travel behaviour and behaviour change, and new mobility data sources, to derive predictive models of travel behaviour under different types of disruption event. We anticipate that these models will reflect facets of choice uncertainty, inherent to these scenarios, that differentiate them from travel behaviours under ‘normal’ conditions. Such models have the promise of making novel contributions to research and practice.

Stage 3 – Agent-based model and scenario exploration

The new behavioural model will be integrated within an agent-based model (ABM), built using the MATSim modelling framework. The researcher will benefit from an existing MATSim transport model of the West Midlands region, but it is anticipated that to ensure smooth integration of these models, further calibration, validation, and development of the model will be required.

Once fully developed, the ABM will allow the testing of different future scenarios, co-developed with TfWM colleagues. The exploration of scenarios will carefully consider their role in appraisal – producing measures of the entire transport system that describe the impacts of different disruption events. These appraisal measures will incorporate dimensions of impact including socioeconomic equity and air quality impacts.

Data Sources

Possible sources of data available to this project include:

  • Wejo Connected Vehicle Trajectories, provided by the ESRC Consumer Data Research Centre
  • Spectus app-derived Origin-Destination data, provided by the ESRC Consumer Data Research Centre
  • Origin-Destination mobile phone data from Connected Places Catapult
  • TfWM data sources including vehicle and cycle counts, travel times, origin-destination flows (from mobile phone data), infrastructure and timetabling data.