Traffic in future cities: network optimisation under new mobility services

Urban congestion in the UK was estimated to cost £30.8bn in 2016, with the UK ranked as 4th most congested developed country1. The CBI 2015 infrastructure report2 headlined "96% of firms concerned about congestion on the road network", and congestion costs are forecast to get worse: "From 2013 to 2030, the annual cost of road congestion is forecast to rise by 63%" according to a recent CEBR report3. The arrival of autonomous vehicles and intelligent mobility services is likely to further increase travel demand, however these advances in technology also open up new ways to optimise the use of existing infrastructure. This is the focus of the proposed project. To develop models and analytical methods to tackle congestion and improve mobility in future cities, and gain understanding of how transport network performance depends on its topology. These findings will enable better strategies to be developed for broader aims of increasing economic productivity, reducing pollution and improving the wellbeing of the urban population.
A substantial volume of academic research concerns the optimal design and management of urban road networks to minimize delays, emissions and fuel consumption. The discrete network design problem (DNDP) is usually associated with road construction (where to add a new link). Technological advances will broaden transport provision and lead to new opportunities for network optimization. Intelligent vehicle routing and V2I (Vehicle-to-Infrastructure) communication will allow the network to be dynamically reconfigured to meet current travel demands. (i) Vehicles could be routed in order to minimize total system travel time; currently each individual seeks to minimize their individual travel time (i.e. user equilibrium), which is not system-optimal. (ii) The direction of road link/traffic lanes can be reversed (i.e. network-wide contraflow) to accommodate network flows and vehicles rerouted in real time. Dynamically optimising the use of existing infrastructure capacity is economical and avoids increasing use of urban space for transport provision. Meanwhile, travellers will select from a variety of mobility services: not only private car and mass public transit, but demand responsive transport and ride sharing services. (iii) Ride sharing has significant potential to reduce congestion (depending on routing, volume of empty trips and the response of travel demand).
Objectives
1. Develop an optimisation framework to solve the DNDP with link/lane reversal (applicable to both user equilibrium and system optimal vehicle routing).
2. Under (1) determine the efficiency gain under different configurations of demand and supply.
3. Solve the traffic assignment problem with ride sharing and develop strategies to maximise efficiency gains (from 1) by incentivising travellers and ridesharing operators to follow system optimal paths and hence improve network-level route flows.
When implementing link/lane reversal, feasible routes need to be maintained for all travellers; the impact of this constraint depends on the network topology (consider grid versus tree network), and will affect (potential) efficiency gains. Infrastructure optimisation across different network topologies is a newly emerging area of work in which Objective 1 is original. This approach establishes the foundation for Objective 2, which will be extended in Objective 3 to demonstrate how ridesharing can be incorporated into transport systems in a way that reduces network-wide congestion and consequently benefits society as a whole.
1 http://inrix.com/scorecard/ (accessed 25/01/2018)
2 http://www.cbi.org.uk/news/infrastructure-survey-2015/infrastructure-survey-2015/ (accessed 25/01/2018)