This article was originally published in French at Alliancy and is reproduced with permission.
Serge Blumental: RTE has a large network, with many kilometers of lines, which was built in big historical waves. We need to absorb a growing technical debt, whilst continuously investing in the network. This is a major challenge in our strategic plan, because if we do not change the size of our renewal policies, we will find ourselves facing an insurmountable financial obstacle. Historically, our industrial assets, such as pylons, conductors, electrical substations, etc., were considered independent from one another, as if they didn’t overlap. However, the demands of network maintenance intersect with human, financial and technical constraints and issues… This is why we have decided to model this reality using a digital twin.
Gabriel Bareux: Our problem is the confrontation of short and long term trade-offs between OPEX and CAPEX. From a technical point of view, it seems obvious, for example, that our pylons need to be repainted regularly to protect them from corrosion and extend their lifespan. But when a company needs to make savings, it tends to reduce these items of expenditure to a short-term vision. And we don’t always know how to find the right balance. With the use of modelling, we were able to demonstrate that doubling or even tripling the effort in terms of painting to prevent corrosion was much more cost-effective in the long run. Alternatively, we were able to determine when it was more cost-effective to simply replace certain classes of pylons. We were able to unfold the possible strategies, with supporting evidence, in order to discuss practical scenarios with the regulator (the Commission for Regulation of Energy, or CRE – ed.).
S.B.: Modelling is a comprehensive and difficult exercise, which requires us to ask questions that we have not necessarily asked ourselves in the past. So yes, it implies a change in our practices, especially as, beyond this project on corrosion prevention, the CRE has asked us to extend this use to other asset management policies. A good example is the maintenance of metal-clad substations. The insulating gases we use can end up leaking out, contributing to an increased greenhouse effect, so it is our responsibility to ensure that our maintenance practices are the most suitable to prevent this.
G.B. In the long term, all the equipment is potentially concerned: all the overhead links and the electrical substations, whether it be the structure, the transformers or the control elements. This overall management policy is interesting precisely because it leads to a cross-reflection. Should we act by whole items? By large units? Should a single technology be replaced everywhere at once? What are the consequences of each choice? We must be able to choose our battles because we are facing one of the most complex global systems ever built by man, as the Academy of Sciences in the United States has pointed out. If we want to succeed in decarbonizing our economies, we must review all these fundamentals in the next 30 years. This is a huge challenge, and one that we had better not get wrong.
S.B. Indeed, we have a public service mission which obliges us to look at more than just the financial impact. We therefore ‘value’ the other aspects of the problem: we assign a high cost to the energy that is not distributed, if the network is faulty, for example. The same goes for safety risk: a cable falling on the ground is a danger to people that we have to quantify. We also quantify the dangers for the environment: greenhouse gases, pollution from the products used, the consequences of a fire, etc. We are increasingly expanding this consequences reference system in all areas.
G.B. Our teams have been carrying out asset management for a long time on the basis of their traditional expertise but using a spreadsheet-type tool. Switching to a digital twin tool means thinking differently about data, being able to understand and challenge the simulations. It is both a change of skills and a change of behavior, of posture. There is a certain amount of apprehension amongst certain experts who are afraid of having their expertise ‘stolen’ by the tool. However, given the subjects, we will always need experts, but they must be trained and then recruited differently.
S.B. The tool generates extremely large volumes of data, in addition to an already rich collection of dashboards. To take full advantage of the simulation, we need to give ourselves the means to exploit it in terms of data science, which implies rather scarce skills. Business expertise on industrial assets alone is not enough. Ownership is an important issue for the coming months, because initially we used the digital twin as an R&D tool. But as soon as we generalize the practice to our entire asset management policy, it will go beyond this framework and affect many experts. The number and variety of users, with more or less advanced uses, is increasing considerably. At this stage, however, the initial feedback is very positive. They are getting the hang of it and the teams have even implemented uses that we had not previously put in place, such as legacy data!
G.B. Industrialisation, that is the generalization of the approach, implies two challenges. Firstly, an IT challenge, that is, moving from R&D to ownership by the IT teams. The other challenge is ownership by the business teams, which requires suitable support. Then, from a data point of view, it should be pointed out that RTE has made great efforts since the beginning of the 2000s to describe its network and its assets in detail. This is our legacy data. It is not perfect because much of it has been entered by hand. It will therefore be necessary to transcribe the data into the tool but above all, to manage imperfections or omissions: to be tolerant of errors, to detect them and to correct them. However, it is not necessary to have nothing but perfect data to run a simulation! We can tolerate some imperfections. Especially since our tool does not only integrate legacy data: there is also data on the skills and availability of the teams, on the strategic importance of the work. We need this rich raw material to fully capitalise on modelling.
After a career as Scientific Director and Vice Chairman of Innovation for the Veolia Group, Michel Morvan co-founded Cosmo Tech (formerly The Cosmo Company) in 2010, a specialist in the modelling of complex systems, whose activity has gradually moved towards simulation digital twins. “I am convinced that we are now in the century of complexity: we see uncertainty and interdependence everywhere. All businesses are faced with a conflicting injunction: build resilience whilst making massive cost cuts. The only way to respond to this is to have visibility on the impact of decisions, particularly using simulation,” emphasizes the co-founder.
The company works with RTE, but also Renault and Michelin on manufacturing and supply chain issues. “Rather than wanting to simplify complexity, we should see it as an opportunity, because it is this granularity in activities that carries an enormous potential for value. Mastering this granularity of action allows us to know how to act precisely in the face of complexity and thus find new opportunities,” he argues. And the results are impressive: “We applied the same digital twin as that used by RTE to manage the obsolescence of robots in Renault’s factories. On an average maintenance plan of 10 million euros per year, the company saw that it could reduce it by 50 to 80% depending on the year in three different plants over the next five years.”