Skip to main content
— Presentation at RaiLille2017 —
May 10, 2017
Photo 1 Presentation scene
RailLille2017 -7th International Conference on Railway Operations Modelling and Analysis- was held in Lille, France, from 4th to 7th in April, 2017. The conference is renowned for railway field and focused on any aspect of railway planning and operation such as timetabling, traffic management, scheduling & maintenance, passenger flow analysis and so on. In the conference, there were about 200 participants, presenters from Europe such as France, Germany and Sweden were the majority, and from Asia such as Japan, China and India were also numerous. We made a presentation titled "A Customer-oriented Rescheduling Simulator for Large-scale Train-service Disruptions" which co-authored with East Japan Railway Company (Photo 1).
Fig. 1 Definitions of passenger dissatisfaction
Fig. 2 System overview of proposed simulator
In our studies, we targeted the Tokyo metropolitan railway network. Its operations are strictly managed and transport capacity is very high. The Tokyo railway service generally operates very smoothly. On the other hand, large-scale train service disruptions are a serious problem. If an unexpected incident occurs, train dispatchers have to reschedule train timetables to reduce passenger dissatisfaction. However, their success depends on the individual skills of the dispatchers. Additionally, they could not consider whether their decisions were reasonable because they could not evaluate the scale of passenger dissatisfaction quantitatively.
For supporting dispatchers, we developed a macroscopic indicator for evaluating passenger dissatisfaction named SCORE which is short for "scale for customer-oriented railway evaluation" [1-3] (Fig. 1). Passenger dissatisfaction is quantified by calculating extra travel time for each passenger during a disruption using trip logs.
In our study, we developed a prototype of rescheduling simulator for dispatchers (Fig. 2). Dispatchers will be able to find the best practice of rescheduling for passengers after a disruption occurred by using the proposed simulator. The simulated SCORE is calculated using the reference data such as travel time on a normal day, the passenger-demand table, and the original rescheduled timetables. Dispatchers can improve rescheduled timetables on the basis of the original rescheduled timetables.
We confirmed that our simulation method achieved accuracy of 85% in 14 actual disruptions. And, we confirmed that the total extra travel time for two million passengers could be calculated in tens of seconds. In the future, we aim for support of rescheduling operation and improvement of simulation accuracy.
(By YAMASHIRO Masao)