Vehicle emissions and grid decarbonisation
The 1983 UK General Election saw the Labour Party manifesto dubbed the longest suicide note in history. The current policy for decarbonising transport in the UK and Europe may be the most complex one. For the policy to work, it is necessary simultaneously to switch the grid to green sources and fundamentally change the relationship between consumers and their cars, in order to balance that new grid. Both are a major challenge, and if either fails, the whole policy fails. If it does go off plan, we may well end up with undesirable cars being powered by a dirty grid, and an unresolved climate change problem. Are industry and government locked in a suicide pact?
We are not used to living in age of electricity rationing, but this is a real prospect as we try to clean our grid with a big switch to renewable energy sources. The UK is already flirting with using its contingencies, even on the existing less intermittent grid, with fewer electric cars and data centres – although at no point so far has the grid come close to shutting down. On 3 December 2024, headroom was nearly eliminated such that a call for rapid reaction contingencies was initiated. On 8 January 2025, power had to be called from Norway to preserve headroom. These are just the first inklings of a problem, and one that applies to many European countries, not just the UK.
The underlying challenge is that we are trying to expand grid capacity to meet rising demand, while at the same time decarbonising it. The chosen primary route to decarbonisation is renewables – specifically wind and solar. These sources have two limitations. First, as they are intermittent, they need accompanying storage to save the surplus peak energy and release it during dark or windless hours. Second, they have relatively low “capacity factors” – the ratio of actual electricity generated in practice compared to the theoretical maximum. Therefore, it is necessary to “oversize” the installed capacity to generate the same electricity as traditional energy sources. Together, to make this approach work, it is necessary to install significant amounts of wind, solar and storage.
The UK’s National Energy System Operator (NESO), which runs the electricity grid, has published a number of scenarios for electricity demand and supply through to 2050, in the context of aiming for net zero . As a measure of the tightness of supply in 2050, even though the installed capacity of wind is forecast to increase by a factor of five and solar by a factor of six compared to 2023, this is not enough to switch off traditional fossil fuel production. Nuclear is also forecast to increase almost four-fold (which would be great for emissions reduction, but would need enormous commitment to achieve), and interconnections to other countries almost three-fold. Still, not enough.
To fulfil the projected 146% increase in annual electricity demand, the vehicle fleet is expected to contribute in two new ways: “demand management” and “vehicle-to-grid storage.” Demand management and its “smart pricing” seek to shift demand to times when there is surplus renewable power. Vehicle-to-grid (V2G) or bi-directional charging allows the grid to suck energy out of your car when the grid needs it. In other words, you will be constrained in when you can afford to charge up, and you might find a lack of charge in your car for your journey. If, for example, only 20% of cars are plugged in at the crucial time, those connected could lose 3 kWh each hour based on NESO projections. Of forecast peak capacity in 2050 of 119 GW, smart pricing reduces demand by 12 GW and V2G could provide 20 GW of power. Therefore, the vehicle fleet is expected to contribute 27% of peak demand to make the numbers add up. This comes at the cost of constraining personal freedom and the inherent attraction of the motor car. On most days, it will be fine, but consider those dark, still, winter Dunkelflaunten when your car will be an expensive brick. This will reduce the utility of a car, and so the willingness of consumers to pay. Fewer cars will be sold, at lower prices, with damage to the industry and personal welfare.
Some, however, would say that such an outcome would be good if it reduces demand for private motoring and leads to a shift to public transport. The bigger problem that remains is that, even with demand management and V2G storage, grid capacity might still fall well short of growing demand. Of the 116 GW of installed capacity in 2023, 36% of this is to be shut down to meet net zero – primary gas and biomass sources. If we take NESO’s “Electric Engagement” scenario where almost the whole fleet is electrified by 2050, 386 GW of installed capacity is needed. In other words, the “clean” part of the grid in 2023 would need to be increased more than five-fold by 2050. 344 GW of new capacity would need be installed that did not exist in 2023. Just 19% of the forecast grid in 2050 was already in place in 2023. Although the UK in particular has made good progress in decarbonising its grid so far, future infrastructure requirements for 2050 are large and risky. If, for example, we fall 25% short of the target for new build-out, it would leave a supply gap of 68 GW in 2050.
At the same time as we face the risk of falling short on supply, demand could rise more quickly than expected. This is not just speculation, as the question is being forced on us by a seismic change since the vehicle electrification policy was enacted: Artificial Intelligence (AI) is taking off in a way that exceeds the expectations of most. As a result, the well-understood increase in electricity demand needed to support a BEV fleet (around 28 GW in 2050 with unmanaged demand) has now been joined by rapidly growing – and somewhat unpredictable – demand from AI. Just one example, as reported in The Guardian recently, is an application submitted for a new data centre in the UK that would “…cause more greenhouse gas emissions than five international airports.” It is forecast to consume 3.7 bn kWh [3.7 TWh] of energy per year when running flat-out, releasing 857,254 tonnes of carbon dioxide (CO2), based on the current average grid mix.
The same NESO scenario as above assumes electricity demand from data centres to be 54 TWh in 2050. One of the more bullish forecasts is from the BloombergNEF, at 3,700 TWh globally. As the UK is approximately 3% of global GDP, that would imply 111 TWh in the UK. This would reflect 39% of 2023 demand and 16% of forecast demand in 2050. If correct, this would create 57 TWh, or almost 7 GW running constantly, of extra demand on top of the Electric Engagement scenario forecast. For comparison, Wood Mackenzie, a consultancy, is already tracking 134 GW of new data centres in the US, which would be 17 GW if pro-rated to the size of the UK. The BloombergNEF projection may, therefore, turn out to be cautious.
So, we can see that persuading customers to buy BEVs is only part of the challenge. Even if we electrify everything, our demand forecasts must be accurate, supply capacity build must happen, and car owners must be willing to engage with behavioural change. If these conditions are not met, we may not have the capacity necessary to meet demand. On plausible scenarios we could be at least 75 GW short, or 19% of the forecast installed capacity in 2050. In this case, what would happen?
The first instinct would be to “manage” demand further. The 75 GW shortfall assumes the maximum use of vehicle smart charging, so that is not an option. Authorities could move to a harder rationing of electricity for motor vehicles, which would be possible by restricting use of public chargers and more aggressive use of V2G storage capacity. It is likely that authorities would prefer to limit motor vehicle use than home heating or electricity, or industrial activities. With remote working now commonplace, driving would be the first activity to be cut, for all but essential purposes. The alternative would be to keep fossil fuel power generation going for longer, which would be politically highly embarrassing.
Despite the embarrassment, it is possible that governments may keep fossil power stations so people could keep driving. In this case, it would be fair to see vehicles as powered by marginal, “dirty” electricity. At present, the marginal CO2 per kilowatt-hour (kWh) of electricity is 350 in the UK, compared to an average carbon intensity of 124 g/kWh. So, almost three times dirtier at the margin. The European Union (EU) marginal rate is around 550 g/kWh, compared to an average of 244 g/kWh. In Poland, the values rise to 880 and 662 g/kWh respectively. This illustrates that the cleaner the average grid becomes, the greater the proportionate uplift at the margin is likely to be. It is worth noting that France’s current grid carbon intensity is 24 g/kWh on average but 510 g/kWh at the margin; even 18 nuclear power plants with 57 reactors is not always enough.
In future, the carbon intensity of the grid at the margin is likely to remain similar to today, at 350 g/kWh. Applying Emissions Analytics’ own real-world testing and decarbonisation modelling, we see the following. The second column covers all the up- and down-stream carbon in making and ultimately disposing of a vehicle, and liquid fuel production. The second column covers the tailpipe CO2 and the same emissions from electricity generation. Each powertrain/grid combination can then be compared over the life of a car compared to the gasoline ICE baseline.
The “average grid” scenario reflects the situation today, where there is sufficient grid capacity to power the new BEVs, but the sources of energy are mixed, including significant fossil fuel gas. The “marginal grid” scenario is similar to NESO’s Electric Engagement scenario, but where capacity growth falls materially short, vehicle-to-grid does not work, or demand growth is even greater than expected. In other words, the new BEVs are being powered entirely by the marginal, fossil fuel energy.
On the current grid mix, BEVs already reduce lifecycle CO2 by 49%, whereas in the EU it is only 32%. In the worst case scenario, having invested so much in electrifying the fleet, the UK might find only a 16% reduction in CO2 emissions. The EU could be in any even worse position, with CO2 rising by 13%, although this is unlikely to happen as the current marginal sources derived from coal would likely have been replaced by gas by 2050.
Which leaves an interesting dilemma. If we push ahead with a best case scenario that gives 85% CO2 reduction thanks to a clean grid, but we fail to make the grid work or demand soars unexpectedly, we could easily end up in a scenario that would be worse than the low risk option of converting the fleet first to plugless full hybrids, which would allow us to bank 29% CO2 reduction quickly and for low cost. Put another way: if we want to convert the fleet to all-electric by 2050, we must be certain that the grid can accommodate such a fleet cleanly.
The optimal strategy, we would suggest, is to push for hybridisation of the fleet simultaneously with grid decarbonisation, and only push on to fully electrified vehicles when the clean grid capacity is secure. This would be a more robust mix of risk and outcome. It would not meet net zero by 2050, but it would reduce delivery risk, and reduce CO2 more quickly in the early years by avoiding the high manufacturing emissions caused by largescale battery production. As readers of many previous newsletters will recall, Emissions Analytics believes that the data points to hybrids – especially full hybrids, with a decent battery size but no plug – being the best way to decarbonise transport for the next decade. After that, fostering technology-neutral competition between rival technologies would be optimal. During the coming decade, investment should be sharply focused on decarbonising the electricity grid, rather than subsidising well-off people to buy expensive (and heavy) pure electric cars.
If we mess this up, we might yet end up living the joke of having to charge up our electric vehicles with a diesel generator. (The AI-generated image above probably cost us 5 grams of CO2 emissions…) Having just recovered from a recent visit to an unnamed low carbon vehicle show that involved entering through unmistakable clouds of diesel fumes from the backup generators running the stands, this is clearly undesirable. At Emissions Analytics’ most recent conference, called Off-Highway Powertrain & Fuels and which we hosted in Chicago, a session stood about these static power sources. Demand is soaring for utility-scale, diesel-fuelled generators, most notably to power data centres to fulfil the already-voracious appetite AI systems have for energy.
Has the suicide note already been signed?
Postscript
We have taken a largely UK and European perspective in this newsletter, but similar arguments are playing out in the USA. For an insightful read from that perspective, we would recommend the article U.S. Energy Policy Undercuts EVs to Make Way for AI by Tammy Klein published recently in Transport Energy Strategies.