Gerhard Kreß is surrounded by trains – both real and virtual. In front of him, on a monitor, is a schematic drawing of a vehicle. A whiteboard looms behind him in the open-plan office of the Siemens Mobility Data Services Center (MDS) that he heads up. The whiteboard is dotted with red and blue formulas and equations that describe what happens during train operations. If Kreß looks out the window, he can see the saw-tooth roofs of the Allach industrial site on the outskirts of Munich where Siemens currently builds its Vectron class locomotives. As of last month, this is also where the locomotives can be serviced and maintained – at the Rail Service Center located just three rail car lengths away. Leading to the vehicle workshop are not only two tracks, but two worlds: the virtual and the real.
Digital Transformation
The data streams from locomotives, high-speed trains, and local trains from Europe and other, non-european countries converge at the MDS. By drawing upon this data, the organisation’s 20 programmers, database experts, and implementation managers have developed a data-driven service for the rail sector which provides real-time train monitoring, forecasting of wear and failure of components, and analysis of complex vehicle problems.
The Siemens Rail Service Center in Allach
“Before a rail vehicle rolls into our Service Center, we already know what needs to be done,” says Kreß. “We did not begin building this group until mid-2014. It was no small thing, because data analysis experts are in great demand.”
The digital transformation is setting the pace for rail technologies. Maintenance traditionally consisted of checking rail vehicles at operating centres on a regular basis, resolving obvious problems, and maintaining machines. Digital technology has opened the door to a new level of service. Remotely or locally collected sensor data, error messages, and log files provide MDS employees with an unprecedented level of detail regarding rail vehicles.
Billions of Data Points
Indeed, digital technologies provide experts with much more than just information on standard variables such as speed, braking behaviour, and mileage. They can also provide information about the behaviour of compressors, the weight of connected rail cars, and the status of automatic control processes. What is more, the quality of the rails, gradients, and slopes, as well as weather conditions during operation are registered along with the operating frequency of trains in rail networks.
Once a train reaches the Service Center, maintenance measures are ready for implementation.
“For the future of the Mobility business, vehicles alone are not the decisive factor,” says Kreß. “For customers, it is about the lifetime costs of vehicles and their efficient use. Success can be achieved only with the help of bundled data from vehicles, the infrastructure, and operations.”
All of this results in a mountain of data. A fleet of 100 rail cars produces about 100 to 200 billion data points every year. And that’s just the beginning. As they analyse this data, Kreß and his team are looking for meaningful patterns. The resulting knowledge can enable the MDS to, for instance, optimize maintenance processes. With gearbox bearings, for example, which are subject to a high level of wear and tear at high speeds, the MDS can predict problems at least three days in advance. Breakdowns are thus avoided, which increases the availability of trains and saves money.
High Speed and Reliability
Behind this approach is a forecasting model developed by the MDS team that analyses mobility systems data. Data analysts first used conventional machine learning algorithms to evaluate the sensor and infrastructure data from a wide variety of trains. This process requires in-depth knowledge of the underlying relationships between systems. Such information can be obtained from train engineers, train manufacturers, and the employees of the new Rail Service Center in Allach.
Just how well this works can be seen in the high-speed rail line that the Spanish National Railways (Renfe) operate between Madrid and Barcelona. Here, Renfe competes with an air route. The train takes two and a half hours compared to a flight time of an hour and twenty minutes. Renfe, however, guarantees train passengers that they will receive a complete refund of their fare if the train is delayed by 15 minutes or more.
To guarantee a high level of reliability, Renfe partenred with Siemens to establish a joint venture that uses advanced data analysis for trains. The result has been only one noteworthy delay, related to technical problems, over the course of around 2,300 trips.
A brake failure that involves error messages, for example, can be normal if the locomotive simultaneously hooks up to a rail car. This kind of knowledge makes it possible to distinguish between important and unimportant factors and recognize causal chains. Thanks to this approach, after just one year Kreß and his team are already able to use forecasting models with a high level of reliability.
From Big Data to Focused Solutions
The Mobility Data Services Center offers an additional benefit: Not only can it draw upon datasets from different rail fleets, but it can also use information about the different conditions under which those fleets operate – whether in Germany, Spain, or Russia. All of this adds up to an information toolbox that can translate into enhanced rail vehicle reliability. This can be an advantage for smaller operators as well, as they can benefit from MDS services to reduce risks.
Kreß says: “Forecasts of breakdowns and wear, error diagnostics, and well-planned maintenance cycles are just the beginning.
“In the future it is conceivable that at the Rail Service Center we will be able to download a vehicle’s complete database, as is now possible with airplanes, in order to review the data for anomalies.”
Data analytics is providing added value to its customers here, leveraged by knowledge about trains and their maintenance acquired over decades – an expertise that is used on a daily basis at the Rail Service Center in Allach. “As an isolated company, MDS would certainly not be as powerful,” says Kress. “And in the end someone would have to pick up some tools and get to work.”
©Siemens AG, Pictures of the Future online magazine