RAID: Monitoring technology for rotating machines

At the CSTJF, the RAID (Remote Assistance Intervention & Diagnostic) team centralizes operating data – vibration, temperature, pressure, speed, flow rate, etc. – from over 33,000 sensors installed on 800 rotating machines: compressors, pumps, gas turbines, diesel engines and electric motors, and so on. Remote monitoring to help predict failures.

 
"Our analysts have over 15 years’ experience. They are often former maintenance engineers with extensive on-site experience and are able to rapidly understand what is happening, thousands of kilometers away from the Béarn region!"

Alexandre Cassaigneau,
RAID team Leader – Centre Scientifique et Technique Jean Feger – CSTJF Pau

With around twenty specialists and experts, the rotating machinery department supports new projects in the Company, in refineries and on oil & gas platforms. Among other things, it helps define the architecture for rotating machines on future sites: sizing, power, number, etc. A task that was extended in 2013 with the creation of the RAID team (up to four people working full time), devoted to monitoring the machines installed on production sites. "A change in strategy", explains Alexandre Cassaigneau, RAID team leader. "This kind of monitoring was routinely outsourced in the past, but TotalEnergies has chosen to keep it in-house so that we can improve the prediction of machine failures and production shutdown". And the results are immediate. A few months later, the implementation of RAID was already improving breakdown predictions and planning for them. "Bringing the service back in-house had several positive impacts: a global approach to equipment using their operating conditions (flow-rate, fluid pressures, etc.), the possibility of comparing suppliers against one another (number of failures, duration, age of the machine when it failed, etc.) and our analysts learned new competencies". Centralizing affiliate data at the CSTJF is the core factor in the revamp of some of our maintenance strategies, now partly predictive, shared by all sites, and in improving the way in which best practices are shared.

Decoding each alert

The software used in RAID collects data from the machines and issues alerts using a prediction algorithm trained on operating logs. When the equipment "goes into the red", the RAID analysts contact the site maintenance team. An inspection operation is triggered and worn parts are changed before failure. "Certain degradations may develop over several months, hence the challenge of predictive maintenance," emphasizes Alexandre Cassaigneau. "With 50 to 200 sensors per machine, we need to identify the risk of failure before it happens, to prevent shutdown and production shortfalls". Today, RAID know-how is part of the Company strategy, and is also being applied to wind farms and combined-cycle power plants, so that the experience gained in the oil & gas sector can be used in TotalEnergies’ new energies sectors.