Data keeps SNCF trains running

Posted: 7 June, 2018

When you're dealing with complex equipment, it's important to get maintenance right. If maintenance is too infrequent, a smoothly running machine can become an expensive equipment failure in seconds. But traditional preventive maintenance, in which equipment is serviced according to a schedule, can be costly. Personnel time, equipment downtime and unnecessary replacement of parts all eat into the bottom line.

There's a better way: Condition-based maintenance. With this strategy, asset managers can use operational data to track the condition of equipment in real time and perform maintenance only when that data indicates wear and tear on the equipment.

When implemented well, a good condition-based maintenance strategy cuts down on unnecessary servicing, reduces the cost of maintenance and improves the lifespan and reliability of equipment. But in order to make a condition-based maintenance strategy work, you need to collect and analyze data about your equipment in an intelligent way.

That's where AVEVA™ PI System™ comes in.

For SNCF Réseau, the state-owned company that oversees the French railway system, PI System has been instrumental in making the shift from reactive and preventive maintenance to a more effective condition-based maintenance strategy. At a 2017 conference, SNCF Réseau's chief data officer, Stéphane de Paris, spoke about the company's successful initiative to roll out condition-based management across the French railway network, using PI System to harness and analyze data coming from a legacy system of equipment sensors in the field.

By using PI System to combine legacy data with powerful new machine learning tools, SNCF Réseau is now looking even further ahead. Recently, SNCF Réseau partnered with Datapred, a French company that specializes in machine learning. Using SNCF Réseau data collected and organized by PI System, Datapred algorithms comb time-series curves for statistical anomalies that might predict equipment failure—even before sensors indicate that the condition of the equipment is degrading.

“The beauty of the system is that it can be used for any type of time series: vibration, temperature, pressure, etc,” said Thomas Oriol, co-founder of Datapred. “Potentially speaking, there's no limit to the type of data that can be incorporated within the system.”