Walking through a railway station, and getting on or off a train at first appear straightforward activities. But travel delays and passenger injuries can result from sub-optimal design of the train, the platforms, or how they meet (the “platform train interface”). While a moment of hesitation deciding which way to walk, or waiting for a crowded train doorway to clear is usually insignificant for an individual traveller, the overall effect of hundreds of similar events can be substantial. The result is delay from unpredictable train departures, or in the worst cases personal accidents when stepping between platform and train.
The RateSetter project will apply data from CCTV on current trains and platforms with novel parallel computing techniques to identify the combinations of platform and train features that set the flow rate of passengers. Items to be studied include vehicle interior layout choices such as positioning of luggage storage, door designs, platform layouts (for example, location of benches and information screens around which people congregate) and platform management (for example, how soon in advance people reach the platform, how people are encouraged to spread out along the platform).
Interactions between platform and train features are key to identifying the best design and operational choices: improved flow rate of people from a train can only be realised if the platform is able to absorb the numbers (and visa versa for people boarding the train). With many thousands of design combinations possible they could not all be tested through building physical mock-ups. Instead, parallel computing using graphics processing units will be employed to rapidly test the options with existing models of people movement, driven by real world behaviours captured from CCTV.
Optimisation techniques including genetic algorithms will be applied to find the strongest combinations of train and platform design. CCTV data will be used to validate people flows for existing fleets, giving confidence in predictive application for novel train and platform designs. Outputs of the project will focus on quick win retro-fit options for improving existing trains and platforms, and more radical options for future stations and fleets.
Read Project Brief