Digitalisation – a key to resilience and cost-effective maintenance

Pierre Dersin, adjunct professor of Operation and Maintenance at Luleå University of Technology, says that planning adequately the maintenance of railways and equipment is a key to success.
“AI and machine learning are important tools for taking the right actions at the right times”, he says.

Pierre Dersin is also involved in the work of AI Factory Railways, AIFR, at Luleå University of Technology where he earlier represented Alstom Digital & Integrated Systems. He is one of those who have given a lecture in the series of webinars organized at the Center for Maintenance and Industrial Service, CMIS, at Luleå University of Technology.

Pierre Dersin discussed models for scheduling maintenance at the right time to keep rail traffic running regularly and punctually. A key concept is system resilience, which can be briefly described as a combination of the ability to absorb “ shocks” (external disturbances), to adapt to them and to recover. Shocks can be caused by a variety of reasons ranging from technical failures, to wilful attacks ( such as cyber-attacks), or extreme climate events.

Keep the funktion

The goal of designing, maintaining and operating for resilience is that not too many interruptions occur, that certain functions keep being fulfilled, perhaps at a degraded level, even though a disturbance problem has occurred, and that the time to return to normal mode is kept short.
There is a trade-off between resilience and capacity: the closer to capacity a system is operated, the less resilient it is, according to Pierre Dersin.
”It is not enough to dimension a system for the normal situation. There must be capacity to take care of the problems that may arise in various areas and at the same time work preventively”.

THE PHM-model

He presented the PHM – Prognostics and Health Management maintenance model. It is about discovering weaknesses or degradations before damage occurs that affects a function.
”Digitalization has fundamentally changed maintenance work”, he says.

He referred to the AI Factory Railway, AIFR, at Luleå University of Technology, where an AI engine is being built. Amounts of data can be entered and the user can ask for single data or contexts, which can then be visualized in for example Virtual Reality-VR.

Digital twin

He also raised the possibility of working with a digital twin.
”An abstract model of a physical object that must be just detailed enough in relation to the set goals, but at the same time contains more data than the physical object itself", he says.
A digital twin also should store useful data, such as history of the maintenance actions that have been performed on the object (for instance, subsets that have been replaced), and the history of its operation profile and health condition. It can then be used for example, for simulations, analyses and prediction of future health, and this can be a powerful support to a PHM strategy.

In summary, Pierre Dersin believes that resilience must be a fundamental property of today's industrial systems. Digitalization is also a key for maintenance and above all preventive maintenance. At the same time, cyber security is a key of importance, he notes.