Electrification is set to be a major factor in the UK’s energy future, but it is not just replacing natural gas for heating. Electricity will also go to new areas of demand, such as EVs and data centres for increased amounts of AI.

These three areas – heat, transport and computing – will see significant growth by 2030.

Each type of new demand will have an impact on the grid in different ways. As such, they will have implications for network planning, generation and storage.

Innovative charging and storage approaches for EVs and complex sustainable supply arrangements for data centres are likely to drive a change in the way the system operates. Meanwhile, heat pumps remain a challenge in the UK – both in terms of roll out and potential grid impact – despite their efficiency and early success elsewhere in northern Europe.

UK power demand has actually fallen since the mid-2000s. Household solar and other decentralised generating assets, alongside improved efficiency of consumer and household goods has driven this change.

Bouncing back

Demand is now rising again, and also changing in nature as the market seeks to redistribute it towards lower cost locations or times. Big consumers are able to choose where they site consumption centres based on availability and price. EVs will be spread fairly evenly, although drivers are increasingly aware of shifting charging to when power is cheapest.

Transmission grid operator, National Grid ESO, models future demand in order to plan the transmission network.

In its Forward Energy Scenario (FES) 10 year outlooks, winter peak – from heating/lighting – has been the main driver of bulk power transfer needs for many years. This is likely to continue, given increased heating demand concentrated in winter from heat electrification. Officially, the government is due to decide on hydrogen’s role in heating in 2026. However, it appears the government has taken the decision.

EV charging will play a key role in managing demand — *Source: Zest*

Only a few years ago, National Grid said that if all gas heating were electrified, the country would need about six times the generating capacity and a massive grid overhaul to meet peak winter demand.

Heat pumps are now more efficient. But heating electrification is still likely to be the biggest potential transition demand factor for the grid, even though it is not the biggest in annual demand terms, that title goes to EVs.

This is because heating demand is concentrated during a few cold midwinter evenings rather than being spread across the year. As such, it therefore puts much greater strain on the grid.

Picking peaks

In its latest FES 2023 report, National Grid expects peak midwinter power demand to rise from 58 GW in 2022 to 69 GW in 2030. It will then reach 87 GW in 2035 and 113 GW in 2050, this would be almost double current levels.

It provides these numbers under its Consumer Transformation scenario, which looks the most likely of the four scenarios modelled given hydrogen’s demise as a domestic heating option. National Grid will release its FES 2024 report in June.

Data centres and AI should have less significant demand in total volume terms than the other two new sources. It will be far steadier throughout the day, as security of supply is of paramount importance for these consumers.

National Grid is also updating its forecasting process to look at other seasonal grid peaks/constraints. For instance, EV charging may jump as summer holidays get underway. Offsetting such a roadtrip peak would be fewer work journeys and higher solar output.

There are differences of opinion over the country’s energy needs. DNV, for instance, expects 28.4 TWh per year of demand from EVs by 2030 in the UK. This will then rise to 62.4 TWh per year by 2035, which would be 22.5 TWh lower than National Grid’s central estimate. It bases this on EV growth of 780% by 2030 and 1981% by 2035.

DNV’s estimates for heat pump demand were much lower than National Grid. It suggested the UK would consume only an additional 7 TWh of power by 2035, with a 735% rise in installations. Looking at current installation rates and efficiency improvements, this could well be a better estimate.

EVs

EV demand is relatively easy to forecast due to the government’s EV fleet sale percentages. This requires EV sales to hit 85% by 2030 and 100% by 2035. Although, of course this could change. Prime Minister Rishi Sunak caused waves in the industry last year when he pushed back the 2030 ban on petrol and diesel cars to 2035.

More EVs will mean a flatter daily demand curve as they snap up the cheaper power during periods of lower demand. There may even come a point when EVs charge when there is more power available – when the wind is blowing or sun shining – to chase the cheapest power. Good Energy’s head of policy Kit Dixon described EV demand as a flexible opportunity. This provides “a valuable source of flexing demand, managed by smart meters to draw power when there is high renewable production”.

Companies like Good Energy and Octopus are already developing products that encourage this. Octopus Intelligent Go charges residential and fleet EVs at will, and also offers a vehicle-to-grid (V2G) service – beginning the process of wiring EV batteries into a national storage system. Planners must also consider home battery storage and solar, especially if prices continue to fall.

Domestic batteries, such as Tesla's Powerwall, will have an impact on grid balancing and EV charging — *Source: Tesla*

Bottlenecks

Dixon said this flexibility meant “generating enough power for the potential 30 million EVs we may have on UK roads should not be a problem by 2050”. But he said there were still challenges related to the public EV charger roll out. “Although we have seen an acceleration, particularly in rapid and ultra rapid chargers in the past 12 months, the bottleneck remains grid connections and infrastructure.”

Robin Heap, CEO of EV charging company Zest disagreed. He suggested the main constraint to the speed of EV infrastructure rollout was “the naturally cautious nature of commercial and public landowners, the time it takes to complete legal and regulatory agreements, with the time it takes to secure new grid connection typically a tertiary issue”.

The fastest way to build EV charging infrastructure is to start where there is already a lot of grid capacity nearby, he said.

The junctions of the strategic road network close to towns and cities, for example, offer abundant power. Energy storage at EV charging hubs may also help with flexibility, although these are rarely located in optimum commercial locations, compared to alternatives closer to renewable generation assets.

Oil demand impact

The switch to EVs will displace considerable volumes of oil from the market, reducing the demand on processing facilities and refineries and the efficiency of the distribution networks. At some point, these may need government support in order to continue functioning, or alternatively may have to put up price significantly to remaining hydrocarbon fuel users.

DNV predicts the UK will see a fall of 130,000 barrels per day of liquid fuel by 2030, rising to 330,000 bpd by 2035. This would be a little over half the current level. Any retail closures or rise in retail fuel prices as scale economies erode would be likely to accelerate the switch to EVs, which are already reaching cost parity with ICE cars in China.

DNV said the lower demand would result in a 17% fall in refinery throughput in the UK by 2030 and a fall of 37% by 2035.

The IEA, in its annual Global Electric Vehicle Outlook released in April, said EVs would displace 6 million bpd of road fuels from global markets by 2030. Last year, the agency predicted it would displace 5mn bpd.

By 2035, EVs will reduce demand by 11mn bpd based on current policies, compared to 700,000 bpd now. The IEA expects global oil demand from road transport to peak around 2025.

Import moves

Under immediate threat appears to be Scotland’s Grangemouth refinery. Petroineos confirmed last year that the complex would only continue until at least spring 2025. The company is a joint venture of PetroChina and Ineos. Next year, the operator will convert the facility into an oil import terminal. This would make Scotland more reliant on imports and be detrimental to the country’s balance of payments.

Scottish and UK governments have been accused of inaction on the issue. However, some work has been carried out on how the UK’s hydrocarbon fuel supply assets would be maintained as the transition progresses. Scottish Labour have urged the UK and Scottish governments to get around the table to discuss the future of Grangemouth.

Data centres

Forecasting AI data centre demand is more difficult than EV demand. Consumption is less certain, as is its geographical spread.

Sally Jacquemin, vice president for power and utilities at AspenTech, said some AI centres would inevitably move nearer sources of cheap, reliable power, such as Scotland. Points of appeal included Innovative Pricing and Power Systems (IPPS) and geographical tariffs that incentivise lower energy costs.

National Grid said there was some potential for locating near renewables and cheaper power. However, other factors were more important, such as proximity to customers, due to communication speeds.

Chris Maclean, head of intermediaries True/Open Energy Market, said the key consideration was availability of grid capacity.

“We are seeing these pockets and concentrations of data centre site investment, in Ireland, UK and Germany, which is reflective of a more modern grid infrastructure where available capacity dictates demand,” he said. These locations were normally just outside big cities or in suburban areas, rather than near renewable generation.

Suburban areas also offer heat recovery potential, for use across local communities, which rural areas cannot provide. They also have better connections to multiple energy sources, in addition to local workers.

“Many data centres are connecting with renewable supplies, but through a PPA, not directly. For example, Google is investing $1bn at a site in Hertfordshire and has aligned with a Scottish offshore wind farm for a PPA – but the generation will be coming in from multiple renewable assets that might not be that close.”

Data centres always need access to reliable supply in order to maintain optimum performance. They cannot rely on one renewable source or on-site generation. Also, he said the scale of data centre demand means any contribution from solar panel roofing is less significant.

Security over emissions

“Data centres are not in the market for demand management and they are keen on price, which can provide a competitive advantage – but not at the expense of security of supply or sustainability. This puts data centres at the forefront of sustainability and security – the eyes of the world are on them. These major new energy consumers need to be bold and clever and manage security well.”

Maclean said it was difficult to estimate additional power requirement from data centres. Often when developers build a new site, they will opt for the largest connection possible. This is even if the facility uses only a fraction of that capacity for several years.

“Data centres are being pro-active, not waiting for AI to arrive and then asking local grid suppliers for additional capacity – they are getting it in place now. We’re seeing huge capacity requests going through, but only 10% will actually be used over the next few years. They are building first, in anticipation of AI demand.” He added that capacity will already have been secured before construction begins.

Once connected and ramped up to full operations, data centre demand is very flat and predictable. However, there can be surges when testing and ramping up that can last a while. These could pose a risk for the grid and suppliers.

Despite their steady appetite, Maclean said data centres are looking at avoiding peak demand times in the UK by switching to batteries, onsite power or some other option. “The next step is for data centres to be a bit more flexible if required.”

Heat pumps

Given the shaky start to the UK’s heat pump roll out and their rapidly improving efficiency, forecasts for heat pump demand vary among analysts. Good Energy’s Dixon said “we found that in an ideal system transition, by 2030, 46% of heat demand would be met through electrification – predominantly heat pumps. But this roll out must be accompanied with extensive retrofitting of housing stock,” which can be problematic.

The practicalities of a heat pump roll remain an obstacle, but their impact on peak winter demand could be moderated by use of more ground source heat pumps and some degree of flexibility, according to Rendesco managing director Alastair Murray.

“Recent research we have conducted with the University of Southampton has shown that heat pump heated homes can see two to three hour windows with their heating switched off with little to no discernible impact on comfort or heat pump performance.”

The increased demand and underlying grid capacity will also depend on what type of heat pump technology is chosen. Air source heat pumps require much more grid capacity per unit than ground source heat pumps, so the latter put less of a strain on the grid.

Murray said air-source heat pumps are especially power hungry on the coldest day of the year when heat is needed the most. Ground source heat pumps can maintain high efficiency even on the coldest days of the year, drawing heat from depth. “Diversifying to ensure that a significant number of homes utilise ground source technology is therefore essential to minimising grid demand,” he said.

Evolving demand

It can be tempting to try and take consolation in the idea of static demand, particularly when notions around distribution and generation are evolving. However, changes in demand will have an impact on systems planning. No one part of the system remains static and changes in one part may have an unexpected impact on others.

EVs are on the rise and heat pumps are clearly going to make progress, as long as political will remains. AI, though, is the unknown. Different sources all make different assumptions, and in all likelihood will be incorrect. AI investors are taking serious steps to ensure they have access to power. While the Googles and Microsofts of this world would prefer to secure renewable energy, that will fall away in favour of reliability.

Just as there is a dynamic interplay of generation and grid, so too is demand in flux. Demand is slower to shift, but long-term predictions are challenging.

Related Content

Getting serious on cybers­ecurity

AI everywhere, all the time

Getting serious on cybersecurity