Alstom is a global leader in the transportation sector. The company offers trains, signaling, maintenance services, as well as integrated transport systems. Among Alstom’s products are the TGV and Eurostar high-speed trains. The company has 105 sites with more than 34,000 employees worldwide. Its net income is €475 mln.
SimPlan AG is the leading German simulation service provider, specializing in the automotive and logistics fields. The company’s revenue is €14,5 mln.
Alstom considers innovation to be crucial in meeting the mobility challenges of the future. Together, with SimPlan, they agreed on developing a digital decision support system for train fleet maintenance management.
This work is a part of the EU OPTIMISED project, financed by the European Commission, as part of the EU H2020 program. OPTIMISED is a large European initiative aiming to develop methods and tools for highly optimized reactive planning across a variety of industrial sectors. Key to the project is simulation and digital twins, built with its help. A digital twin is a virtual replica of a physical system and its operations, as they operate in real life. This type of simulation model can be continuously updated from multiple data sources and change its state to represent its physical counterpart.
Alstom maintains the entire Pendolino train fleet on the busy and constrained West Coast Main Line (WCML) in the United Kingdom. With 56 trainsets to be maintained and five maintenance depots, the company has to take many aspects into account when scheduling and managing maintenance:
- Daily operating requirements for the routes and timetables regarding the trainsets and capacities needed.
- Maintenance regimes: frequency and parameters (such as time or mileage) for trainset inspection and maintenance.
- Corrective maintenance in case of accident or failure.
- Maintenance capacity: whether a depot has enough resources for maintenance or repair.
Trains serviced earlier than needed cause unnecessary expense for the maintenance company, while late maintenance can result in failure and additional costly repair. So, a comprehensive digital tool was required to help manage maintenance effectively.
As there are many parameters to be considered, simulation is needed. But simple simulation with fixed data is insufficient. The reason for this is that the railway situation is very changeable, despite there being a fixed train timetable, and it is very hard to predict train locations, even a few days ahead. Using up-to-date data would enable deeper insight, and this led the developers to build system's digital twin. With daily operational updates, it became possible to represent the system accurately.
AnyLogic transport simulation and planning software enables the use of the most suitable modeling method for a simulation or even the use of several methods together. For this model, the developers chose an agent-based modeling approach, which made it possible to capture the whole railway network and operations:
- The fleet
- Depots and stations
- Maintenance regimes
- Diagrams that define fleet schedule
AnyLogic also enabled the developers to handle data from different sources without changing format. System data on fleet, stations, depots, and their constraints are provided in Excel, while the trainset scheduling diagrams are CSV-files and assigned on a daily basis.
The maintenance scheduler Alstom usually uses is based on a heuristic scheduling algorithm and the developers embedded it inside AnyLogic railway simulation software. This provides a big advantage because connecting the simulation and the scheduler directly means they can be rerun together for quicker results, whenever needed.
The model has an interactive and user-friendly AnyLogic interface. The GIS functionality in AnyLogic makes it possible to display and manage GIS maps in the model. Using this functionality, the developers visualized railway fleet operations using data from OpenRailwayMap. On this map, users can see all fleet operations. Moreover, it is possible to click on any item and get comprehensive information about it. For a trainset, there are:
- Statistics on its cumulative working hours.
- Details of preventive and corrective maintenance, both completed, due, and scheduled, and the depots involved.
- Total amount of time the train can be out of work for maintenance due to schedule.
Being Java-based, AnyLogic also allowed the developers to create custom Java extensions and a freely distributable standalone application for railroad fleet simulation and optimization — a feature which helped engineers present the model to executives.
Numerous further developments of the simulation model are planned. For example, fitting trainsets with detectors to send data to the model and help make it represent reality more closely. The scheduler is also to be upgraded with probabilistic methods and machine-learning features for predictive rail scheduling and further optimization of scheduling policies.
The digital twin represents the operations of the entire WCML fleet. It enables its users to save on unnecessary railway maintenance expenses by finding an optimum solution for the given constraints. The user can:
- Understand the system performance within given parameters and find bottlenecks.
- Explore different ways to service trains more cost effectively (altering train fleet maintenance regimes, scheduling strategies, depot capacities), fast, and safely in a digital environment.
- Compare scenarios, evaluate KPIs, and make informed decisions.
In case of any emergencies or unplanned events, Alstom can quickly find a new and effective solution by changing the input data. It is also possible to anticipate possible events and find solutions beforehand by running various what-if scenarios.
If some global changes are proposed by the customer (new timetable, additional trains or routes), the maintenance company can check whether they affect maintenance and propose new solutions. Additionally, the model is a good illustrative tool when presenting to the customer.
The rail network digital twin is a valuable railroad simulation and decision-support tool for other stages of fleet maintenance, including:
- Taking part in bids and tenders, the company can make reliable estimates with data and simulations to support proposals — the model is a powerful visual tool for communication.
- During the design and engineering phase, the maintenance company can be flexible with project changes and consider limitations.
- The company can make forecasts based on the model outputs and make decisions considering project end-of-life phase.
The investment in a rail network digital twin has proven very useful for decision-making, both in the present and the long term.
Our white paper, An Introduction to Digital Twin Development, contains further case studies that help demonstrate the development of digital twins and their benefits — download.