Accelerate technological developments, optimize the implementation of production fleets, drive supply and demand… In many areas, energy players are relying on AI to facilitate the energy transition.
In some areas, AI could be used to accelerate research and development by analysing large volumes of data and work
In a decade, digital technology has spread to all spheres of society. The advent of industry giants, the so-called Gafam, and the huge investments that followed, accelerated work on artificial intelligence (AI). Now, some are predicting the next industrial revolution with computers capable of sorting, but also analyzing, interpreting and even deciding, thanks to the masses of data produced by ubiquitous connected objects.
In the energy sector, artificial intelligence could accelerate the transition from centralized production, with a few dozen facilities, to a fleet of thousands of decentralized facilities. From production to distribution to trading, AI has already invested in energy for the past four to five years. It brings new actors and new ways of working. “Energy players alone would not be able to offer these new tools,”explains Flavien Vottero, director of strategic and economic studies for the Xerfi Group. But energy companies must not be dispossessed, there is a balance to be struck.”
Accelerating technological developments
IFP New Energy (Ifpen) has chosen to work with the National Research Institute for Digital Science and Technology (Inria). “We are in a race for complexity, we are going further and further. Digital technology will take on an increasingly important place and this place will grow again tomorrow, with quantum computing. This requires new skills and new trades,” says Dominique Humeau, research director at Ifpen.
Both partners are focusing their research on the contributions of AI in the energy transition. “The economic stakes are high. It is a question of substituting traditional energies developed for decades, with a strong return of experiences. By going further in simulation, the aim is to develop new technologies already optimized and thus reduce costs,”explains the researcher.
One of their areas of research is to optimize floating wind turbines and increase their lifespan. “This technology will be able to capture more wind, but it is also more complex than offshore wind turbines. With floats, structures get tired faster.” The data collected from sensors will allow to model the operation of the machines and to continuously estimate their damage, but also the correct orientation of the blades. Ai could save a few years of research and testing and allow optimized technologies, with fewer materials, to be brought to market at a lower cost.
The concept of “digital twins” already exists in industry and in the energy sector. It is a question of creating the digital double of a production facility, and modeling its operation to anticipate wear, plan maintenance… “This lowers the cost of electricity generated by extending the life of the facilities,” explains Dominique Humeau.
Optimizing the implementation and integration of renewables
AI is also used to improve the implementation of production projects.. “Intensive AI calculations have traditionally been used in the exploration and development of new hydrocarbon production sites. Today, small projects are being developed to optimize the implementation of solar and wind projects according to the wind deposits, the sun… saysFlavien Vottero, Xerfi’s director of studies. But in France, demand is still reduced.
The simulations also optimize the location of wind turbines within a park, to limit wake effects. “We are working to accelerate complex simulations, which mobilize phenomenal computational means, by simplifying them. But the goal is to keep a high accuracy, adapting the no calculation to the need. This allows us to develop tools that are less energy-intensive,” explains Dominique Humeau.
The AI is also involved in the management of networks and smart grids: remote monitoring, analysis of needs andforecasts, integration of renewable energies… Research and development projects are long and often require several years of experimentation.
Share knowledge and research
Intensive AI calculations have traditionally been used in the exploration and development of new hydrocarbon production sites.
Flavien Vottero, Director of Studies at Xerfi
In some areas, AI could be used to accelerate research and development by analyzing large volumes of data and work. Ifpen and Inria are working on molecular simulation and machine learning in the field of biomass, in order to discover new catalysts. “Many researchers work on the same tools but without sharing. The idea is to improve existing data capitalization technologies. It would save time for research.”
While search engines have now become very effective on natural languages, the results are not there as soon as we switch to the scientific field. “The goal is to create a tool that allows us to see all the advances in a specific field in the scientific literature. The opening of knowledge is in tune with the times. Most french and European-funded research projects include publication of the results,” says Dominique Humeau.
Adapting offers and offering new services
Downstream, AI also takes a place: “AI tools are widely used in the field of trading; they allow historical modelling and price projections based on changes in consumption, production jumps…, explains Flavien Vottero. These tools are quite advanced but they found their limit during the Covid-19 crisis,” adds the director of studies.. When the unpredictable happens, even the AI loses its Latin…
The last link in the chain, energy suppliers are also interested in these technologies to improve their customer relationships and marketing strategy. A better knowledge of customers allows them to offer suitableoffers. “Total Direct Energy, for example, does targeted advertising. These tools are cross-cutting to all economic sectors; they optimize costs,” notes Flavien Vottero.
In the longer term, AI will enable them to offer intelligent energy efficiency services: identifying energy-saving levers, controlling consumption, optimizing production, etc. “This is the most complex area to deploy. There are no plans for a massive deployment until 2025,” says Xerfi’s expert. This area could also be the juiciest: many players position themselves on connected objects, starting with the Gafam.