Real-Time Rail Scheduling
Our team worked closely with the company’s rail business unit to learn user expectations and gather business requirements. Then, we developed and deployed the train scheduling system in multiple phases. The first phase of the system went live within one year, and we are continuing to add new capabilities every three and six months since then, and each capability bringing additional value. At the heart of the train scheduling system was a simulation-guided optimization algorithm that simulated one week of train movements within one minute of computational time. It identified numerous scheduling alternatives, evaluated these with respect to a user-defined objective function and recommended the best alternative.
Our end-to-end optimization algorithm considered production targets (demand), the mine plan (supply) and resource constraints, such as railcars and end-facilities, to generate complete train departure schedules. We packaged our algorithm into an interactive decision support system with a graphical user interface that was easy to learn and use. Multiple users with different privileges were able to work in parallel, and each user could create multiple scenarios. We displayed the train schedule using tables, charts and an electronic train graph (ETG), which presented schedules in an intuitive way. Using the ETG, users could add train dwell time, maintenance events, speed restrictions, switch clamps and crew changeover.
- Our innovative algorithms integrated the company’s data sources to create accurate and executable train schedules that showed the potential for adding trains with new revenue of several hundred million dollars annually. The project team won an award from company executives for being one of the most value-adding initiatives of 2017.
- Train schedules generated by RailMAX had lower cycle times, leading to more efficient use of rail infrastructure and increased throughput of the rail network.
- The system enabled smart train scheduling on the day of operation and up to a week in advance, eliminating the need for manually-generated short-term schedules. Network coordinators quickly managed deviations by re-optimizing schedules after changes occurred in the rail network.