E-FWDThe E-FWD logo.


AI to tackle the transmission challenge

The transmission and distribution grids face a mammoth task. They must fundamentally change in order to cope with more and more decentralised intermittent generation, alongside increased demand. Could AI transform this problem?

To give a sense of the scale, National Grid ESO plans to spend £54 billion on the grid by 2030 – the same as it has spent over the last 30 years. This is already posing a challenge for both manpower and finance, but transmission constraints are already delaying various projects around the country.

However, there may be some shortcuts. AI and clever market design are likely to be able to increase the grid’s changing task. Using these tools can reduce the amount of new transmission and distribution capacity required, cutting expenditure and relieving some of the existing bottlenecks in the system. Cash flow constraints and planning delays often cause these.

AI in the nick of time

Just as the complexity and scale of the required energy transition grid investment is getting out of hand, technology appears to be coming up with some solutions.

AI, in particular, is able to play a crucial role in alleviating the pressure by maximising the grid’s potential. Simon Harvey, group vice president of energy and commodities at Publicis Sapient, said measures were needed to stop constraints impeding the energy transition. These included delayed connections and renewable load shedding.

“AI is poised to revolutionise the power utilities industry and specifically the transmission and distribution networks, by enhancing asset optimisation, efficiency and sustainability,” he said.

AI and technology are changing the management of electrical grids, Davi Carvalho Mota, partner for long life infrastructure at Actis, said.

“From precise demand forecasting to optimising power generation mixes, AI algorithms can analyse vast data to predict and meet electricity needs efficiently. AI enables demand response programmes, smart grid management and predictive maintenance, reducing costs and downtime.”

Taco Engelaar, managing director at energy infrastructure specialists Neara, has noted there is idle capacity that can be used.

In some cases, AI and digital modelling can double the available capacity. “Harnessing these tools to locate latent capacity where more renewable energy could be safely connected to the grid is a more immediate, and cheaper solution to transmission constraints and [to] meet evolving demand,” Engelaar said.

Scotland has extraordinary renewable potential, but the grid has struggled to deliver power to demand.
National Grid

Smart grid management

A primary use for AI is to improve load balancing. This involves dynamically adjusting supply based on real-time demand, which will reduce the risk of blackouts, overloads and load shedding of green power.

Harvey noted AI was able to leverage real time and historic asset data from sensors, imagery, consumption and customer data from smart meters. Using this information, it can maximise power network availability and facilitate the integration of renewable energy sources.

AI can enable an intelligent grid, Schneider Electric chief innovation officer Nadège Petit said. This can “bring in more sources of power from the edge – helping ease the tension between limited energy supply and increased demand, and reducing reliance on expensive and polluting sources to deal with demand peaks.”

Mota noted AI’s ability to forecast intermittent sources and co-ordinate with storage systems to enhance renewable energy integration.

VPP plans

“Moreover, AI can orchestrate distributed energy resources like rooftop solar into virtual power plants [VPPs], improving grid resilience. As we transition to sustainable energy, AI’s potential to intelligently balance supply and demand is game-changing and this obviously has a significant impact on transmission networks.”

Petit agreed. She noted that AI could knit together grid-connected assets to create more efficient and flexible capacity.

“AI-powered [VPPs] can aggregate and optimise large portfolios of distributed energy resources (DERs), from EVs and battery storage, to solar panels and smart loads.” By using algorithms to account for each specific characteristic, AI can maximise the overall utilisation of DERs.

This, in turn, reduces the need for balancing with inefficient and polluting sources, such as gas peaking plants. By being smarter, the grid can be cleaner.

AI can also have a part to play in price prediction. This should dampen volatility and better match supply and demand.

“These models use deep learning to notice patterns in energy pricing and predict them on the day ahead, intraday, imbalance and balancing markets,” Storelectric managing director Tallat Azad said.

“Together with large-scale storage, this technology could be used to shield energy companies from price instability caused by the oncoming green transition and enable a stable, cost-effective future grid.”

Balancing mechanism reform

TLT Solicitors partner Michelle Sally said the system operator was using AI to reform the balancing mechanism. It has the “aim of overcoming the issue of ‘skip rates’”. Outdated control room processes have struggled to use batteries to balance the grid, despite often being ahead in the merit order.

She said operators can use AI to co-ordinate demand response programmes. This can incentivise users to reduce or shift their energy use during peak times. On the demand forecasting side, AI can help based on historical data, weather patterns and usage trends, which improves generation efficiency as well as distribution planning.

“In terms of grid resilience and recovery, AI can be utilised to restore grid operations after a disruption by managing resources and guiding repair efforts.”

Sally said there was likely to be guidance soon on deploying AI in the energy sector. The government has decided “to defer to the regulators, rather than introduce an AI specific piece of legislation as done so by Europe, with the AI Act”, she noted.

AI’s ability to enable a centralised dispatch mechanism also interested the industry, TLT’s Juliet Stradling said. The Review of Energy Market Arrangements (REMA) is working on this.

AI can also help predictive maintenance. By examining data from GIS, asset management and maintenance platforms and drone based aerial imagery, these tools can help predict and prevent failures. This reduces maintenance costs and improves plant efficiency.

Utility data for AI LLMs

Utilities can use their extensive historical data to train AI Large Language Models (LLMs). This enables them to predict future behaviour, shifting from rigid maintenance schedules to needs-based strategies, according to Harvey. This increases workforce efficiency and reduces operational costs.

“Power utility companies need to adopt a digital strategy that leverages AI, analytics and their wealth of data (integrated and enhanced with cross-industry open data sharing) to adapt to the evolving requirements of renewable energy,” he said.

Satellites can also feed data into the AI, Damian Lewis, market development manager at Viasat said. This can be combined with a local “internet of things” (IoT) and provide network-wide surveillance of key assets in real time. This provides better data, but also faster reactions to changes in supply and demand.

“As power demand increases and new sources of energy come onstream, the monitoring and control of critical assets will be essential to preventing outages – satellite monitoring will enable such rapid responses,” Lewis said.

Harvey went on note the use of AI to analyse use patterns. This can offer “personalised energy-saving recommendations and, more importantly, personalised tariffs, based upon time and location”. As such, homes and businesses can become smarter.

Ambient balancing

Electrical Grid Monitoring (EGM) CEO Amir Cohen said network operators have been overly cautious in the setting of grid constraints. Setting incorrect and inflexible calculations of thermal capacity leaves the grid to fall short.

AI gives the chances to track all the environmental, physical and electrical factors influencing grid performance, he said. This could dramatically accelerate electrification of the economy without excessive new power grid construction.

“One solution is to create smarter grids that monitor factors such as temperature or windspeed to help take advantage of cooler conditions to safely increase their thermal capacity, with tests already showing 18% gains. AI innovations could help predict opportunities to safely scale up power grid capacity across different seasons or weather conditions. Similar systems can spot opportunities to boost network capacity by sharing loads between parallel feeder lines,” said Cohen.

Weather prediction

Schneider Electric’s Petit pointed to AI for its forecasting ability, with a more accurate prediction of energy generation from sources such as solar and wind. Better visibility on this front would give grid operators more tools to manage intermittency and optimise energy storage.

Andy Willis, the founder of Kona Energy, agreed. Better weather data monitoring and the ability to send quicker responses to power stations would enhance the ability to balance the grid in a cost-effective way, he said.

Increasingly, AI will be at the heart of the modern electrical grid. This will develop predictive models of energy production and consumption, while also managing distributed resources.

This should improve efficiency while reducing the need to continuously add more generation and network capacity. The role for the grid is evolving.

How the grid functions should also evolve. There is a need to expand capacity. But AI and technology provide a means of performing better, rather than just bigger.

Updated on June 5 to correct comments from Viasat’s Damian Lewis.

Related Content