Artificial intelligence: the future of the electricity sector?

Energy storage and demand management…

Now that energy storage technologies are coming close to commercial reality, decades of work should result in artificial intelligence (AI) emerging as the third key technology in the transformation of the electricity sector.  Combined with scalable generation and storage, it will blur the distinction between suppliers and consumers, with excess local generation being fed into the grid so that entities from individual homeowners to business and municipalities will become “producer-consumers” or “prosumers”.  Demand management systems will also have a role to play.  The introduction of multiple players of widely varying consumption and production patterns connecting into a single nationwide grid is impossible until we have software able to predict and manage energy flows to ensure that supply and demand balance at all times.

There are obvious drivers also for energy storage at small scale, particularly for remote locations.  Apart from the potential for autonomy, and the ability to smooth draw from the grid (avoiding or at least reducing demand-based charges), local storage could relieve grid congestion and add flexibility to power generation requirements, potentially improving network stability.

… and microgrids

Many incentives point towards the deployment of microgrids as the next stage – peer-to-peer energy networks that distribute electricity in a small geographic area, usually supplementing, staying connected and continuing to rely on the central power grid but also accessing generating units located at or near their point of use that rely on locally available fuels, especially locally available renewable energy resources.  Microgrids can improve reliability (fewer wires), environmental sustainability, security (no single point of failure) and local economic development and jobs.  Possible business models include third party leasing companies and individuals working with the utilities, or a regulatory shift enabling the existing utilities to include residential solar in their rate base.

… will require the active management which only AI can provide

Only AI will be able to deliver the active management necessary.  Balancing grids, negotiating joint actions to enable self-healing of networks in response to a fault or a hack, demand management and assessing reliability of the production and consumption figures supplied by ‘prosumers’, to name a few areas, will all require real-time forecasting, monitoring and decision taking.  But more than that, the systems running many of those components will each need to learn the detailed behaviour of their own patch: at what times, for instance, is a given household’s electric vehicle (potentially both a form of transport and a temporary energy storage unit) a load on the grid or a supply into the grid? Even at grid level, intelligent networks will be needed to track commercial, industrial and municipal energy usage and automatically charge/ discharge energy from storage to shave peak demand.

Software so closely integrated with the physical networks and generation/ use endpoints will easily meet the ‘technical effect’ threshold set under European patent law for patent eligibility, providing a legal monopoly to protect the research investment underlying intelligent management systems.  Standardisation of some aspects will be necessary to minimise costs in the long run, and work is already in progress, such as the SunSpec Energy Storage Model Specification, or “MESA-Device,” being launched by the MESA Standards Alliance in the USA, whose participants are (apart from Alstom) mainly US entities. This lays out a standardized approach to integrating batteries, inverters, software control systems, and the interface that communicates the data to the outside world.  But for the time being integration of energy storage into other grid components is being approached case by case, resulting in competition not only between suppliers but also between technologies.

An early entrant

A good illustration is an early entrant in this field, Swiss-German firm Alpiq, which in 2014 launched an intelligent system called GridSense which aims to imperceptibly steer domestic or commercial energy consumption of the equipment into which it is integrated (such as boilers, heat pumps, charging stations for electric vehicles) based on measurements such as grid load, weather forecasts and electricity charges.  So far Alpiq appear only to have filed one patent application, in the field of electric vehicles, but of a breadth likely to affect many other businesses thinking of energy management.  It claims a method for programming energy flow in an accumulator of an electric vehicle, by recording information on past usages of the accumulator locally at a point of connection to a grid; locally estimating future usage of the accumulator, taking in input the information on past usages; and locally programming the energy flow between the grid and the accumulator, which may be in either direction, on the basis of the estimated information.  Instead of always charging the accumulator up to 100%, the invention attempts to estimate (based on the charging operations history) how much power is effectively needed for the vehicle’s next outing, using machine learning and data mining techniques.  Alpiq claim that this may lead to improved performance as the charger will not attempt to charge the accumulator up to 100% when just a small amount of energy is needed for the next trip.

Alpiq are unlikely to be the only ones looking to patent AI techniques for managing future electricity flows.  This is going to be an exciting area for energy suppliers, users and traders for some time to come.

Lorna Brazell, Partner, Osborne Clarke

Sign up to our newsletter

Meet our experts