Rail cooperative research centre
-
Development of an integrated train scheduling system for rail networks
-
Development of decision support tools for assessing network performance and reliability
About Rail CRC
The Centre for Industrial and Applied Mathematics is a key research centre in the Rail CRC, formed in 2001. UniSA is one of the several major university partners.
Phil Howlett is the leader for Theme 3: (Optimal Traffic Control and
Scheduling), one of 5 major research themes. There are 4 main projects within
Theme 3 led by Phil Howlett, David Panton, Kevin White and Michael Taylor.
The aim of the theme is to develop optimal traffic control and scheduling
systems that will improve rail productivity by developing good network
infrastructure, good schedules, good recovery techniques, good driving and good
communications. This will be achieved through four major projects:
Development of an In-Cab advice system for improving timekeeping and reducing energy consumption on long-haul trains
(B Benjamin, J Boland, P Howlett, P Pudney, R Raicu and A Torokhti)
The aim of this project is to develop an "in-cab" system that will provide
driving advice that will help the driver achieve the required schedule with
minimum energy consumption.
The system uses the Global Positioning System (GPS) to monitor the progress of the train, and uses these position and speed measurements to calculate key train parameters. These parameters are then used to predict the performance of the train, to calculate an optimal driving strategy to the next target, and to display driving advice to the train driver.
The method for calculating an optimal driving strategy was developed as part of a previous SPIRT grant with TMG International (1999-2001). An optimal control strategy requires the driver to switch between four simple driving modes at various points along the journey. The precise switching points depend on the track elevation, curvature and speed limits, the performance of the train, and on whether the train is ahead of or behind schedule. The strategy saves energy by minimising unnecessary acceleration and braking.
Dynamic rescheduling on long-haul rail networks
(J Filar, S Mackenzie, P Pudney and K White)
Trains on a long-haul network frequently stray from any predetermined plan. When a train is delayed, the effects ripple through the network. These perturbations often require an updated plan that will allow trains to move forward from their current positions in an optimal way.
This project will develop a real-time dynamic rescheduling system that will help train controllers keep the network running smoothly. The system will take into account operational constraints such as speed limits, track maintenance requirements and safe working restrictions. The outcome will be a system that monitors the movement of trains on the network and continually revises the network operating plan to recover, in an optimal way, from disruptions to the planned schedule.
The dynamic rescheduling project extends work done as part of an ARC Linkage grant with TMG International (1999-2001).
Development of an integrated train scheduling system for rail networks
(A Albrecht, G Eitzen and D Panton)
Rail plays a significant role in Australian freight and passenger transportation. Traditionally rail track operators have constructed train timetables by focussing on satisfying customer demand factors and increasing profit. Crew rosters and track maintenance schedules are generally considered once train paths have been determined.
The lack of integration involved in planning these activities means that even if each schedule is optimal in some sense, it is unlikely that, taken from a global view, the schedules will be ideal. The process of interleaving independently created schedules may also reduce their reliability, resulting in timetables that are sensitive to unforeseen events such as delays and breakdowns.
The project aims to develop scheduling tools for rail networks that can be used in both planning and operational environments. The researchers within the project are considering both traditional scheduling formulations and alternative heuristic procedures.
Development of decision support tools for assessing network performance and reliability
(D Lee, R Raicu and M Taylor)
This project will use the Schedulemiser system developed by Scott Mackenzie, David Lee and Peter Pudney.
This project will develop the tools necessary to analyse and assess existing rail networks. Proposed infrastructure improvements, using congestion pricing models, will be planned and evaluated. The outcome will be a collection of high-level tools for planning the long-term development of rail networks.