To speed up its digital moulting, SNCF focuses on industrial Internet. Today, 400 employees of the transport group work on some fifty IoT projects at different stages of development. With the Internet of objects, the group intends to improve the security of its network, the quality of its service and generate significant productivity gains. Here's how.
SNCF places its sights on the Internet of objects. "In our company, IoT is the main lever for performance and efficiency. The true technological revolution is the IoT," said Thursday, May 18, 2017, Guillaume Pepy, the president of the group, at a press conference devoted to the digital transformation of the company, to which the group will spend an additional 900 million euros over the next three years.
400 people dedicated to IoT
"Today, 400 people at the SNCF are working in the IoT, representing 50 teams in different departments of the group," said Benoît Tiers, CEO of e-SNCF, a new entity of the group bringing together the 4,000 employees dedicated to digital and information systems. Today, SNCF counts five industrial Internet projects in the industrialization phase, 12 in pre-industrialization and 30 in the experimental phase. "And 11 of our technocentres are engaged in the transformation of the plant of the future," adds Benoît Tiers.
Of the five IoT projects in the industrialization phase, one finds in particular a system of connected coupons which makes it possible to replace the mercury thermometers. "Before we manually retrieved the data by touring the hot channels during summer, we are now able to measure the temperature remotely with the deployment of 500 connected coupons, which improves the punctuality of our trains, as we reduce their speed only on the portion where the sensors send an alert to indicate that the temperature is above 45 degrees, "explains Claude Solard, Deputy General Manager Security, Innovation & Industrial Performance at SNCF Network.
Combine data to accelerate predictive maintenance
Another example is the deployment of ground cameras to monitor the state of health of pantographs, which are connected to the catenaries and which allow running of the trains. "This equipment falls down very badly but when there is one, it is a disaster, so we carried out periodic checks," explains Claude Solard. The challenge here is to go through a logic of predictive maintenance. The cameras are thus connected and carry self-learning image processing software capable of detecting tiny defects. The system was developed by the start-up GST, based in Dijon. "Six stations have been equipped with these cameras and SNCF Réseau plans to deploy 40 of them by mid-2018," Claude Solard said.
The acceleration of predictive maintenance operations will depend to a large extent on the SNCF's ability to cross its data from a variety of sources. Today, the company works on data from connected objects, employee smartphones or measuring trains, but these data are located in silos. The crossing of this information should enable the teams to move from a descriptive model to a prescriptive model. "We want to be able to predict where we need to intervene and what is the best intervention to achieve, whether it is maintenance or regeneration," explains Benoît Tiers.