Nicolas Charton – AI in Energy: numerous outlets, yet to be achieved

Artificial Intelligence (AI) is a concept that encompasses an array of technologies that aim to mimic functions associated with human intelligence, such as “reasoning” and “learning”, in order to predict, classify or group together data. The technology is not new – in the late 1990s, the Deep Blue supercomputer put an end to humanity’s supremacy in the game of chess.

Author

Nicolas Charton, Managing Director – E-CUBE Strategy Consultants

n

About

Artificial Intelligence opportunities in the energy sector

AI in Energy: numerous outlets, yet to be achieved 

Artificial Intelligence (AI) is a concept that encompasses an array of technologies that aim to mimic functions associated with human intelligence, such as “reasoning” and “learning”, in order to predict, classify or group together data. The technology is not new – in the late 1990s, the Deep Blue supercomputer put an end to humanity’s supremacy in the game of chess. However, it is experiencing rapid acceleration made possible by the democratisation of significant computing power and the advent of big data. Today, it is being used in different practical applications such as image recognition, voice assistants (Alexa, Siri) and the exponential improvement in translation tools (DeepL, Google Translate).

AI is developing rapidly in different industries (transport, manufacturing, pharmaceuticals, banking, etc.). It is driven in the short term by the possibility of optimising existing systems at a lower cost (e.g. improving transport flows), but it is its ability to revolutionise an industry that makes it a major topic for executives (e.g. autonomous vehicles in transport).

The energy sector is also witnessing the emergence of the first practical applications of AI, particularly in forecasting consumption, production or prices on the wholesale markets.

However, the energy sector is not the most straightforward field of application for this technology. AI is very data intensive and is particularly effective when it can be put to work on large amounts of it (on a very large number of cases or very frequently). The energy sector, on the other hand, only handles comparatively small amounts of data – historically, it has focused on centralised assets for which the “human intelligence cost” remains marginal.

This situation is changing, obviously. The use of smart meters is becoming widespread, and almost all new consumers/producers (inverters, batteries, electric vehicles, etc.) are connected. In addition, the gas, electricity and heat network infrastructures are rapidly becoming decentralised, and the scalability of decision-making and optimisation processes is a crucial factor.

 AI will undoubtedly be built on the following dynamics in the energy sector: 

  • Predictive maintenance: Consolidated Edison and Columbia University have developed a model that identifies high-risk components in the New York power grid. A full 60% of the subsequent failures (within the year) were caused by the 15% of components classified as most at risk by the model;
  • Automated advice, energy optimisation and control: Energiency offers a solution for optimising the consumption of industrial processes based on artificial intelligence. It predicts consumption very accurately in real time and identifies possible setting errors or failures;
  • Personalised customer experience (one-to-one marketing): As in other sectors, AI solutions can predict which customers are likely to change supplier or can be used to manage 50% to 80% of the customer relationship. With the prospect of liberalisation, these tools could become essential;
  • Intelligent deployment of decentralised infrastructures – prosumer groups, biogas plants, charging stations, microgrids etc.: Gas and electricity providers are studying the possibility of using AI to identify and optimise complex deployments (for example, plans for a biogas plant must integrate input potentials, transport costs, network capacities, heating needs, etc.).

With few exceptions, these models have yet to be proved and then generalised on a large scale. They are also more industry optimisations than disruptions. In the energy sector, efforts in digitisation appear necessary before AI can develop massively. Once this step has been taken, AI opportunities could abound in the energy sector.

About the author

Nicolas Charton is managing director of the Lausanne office of E-CUBE Strategy Consultants, which specialises in the energy and mobility sectors. An engineer with expertise in industrial organisation, he advises the management of major Swiss and European energy companies, and public bodies such as cantons.