Unless you’ve been on sabbatical to the Moon in recent months, you’ll be aware that there has been lots of talk and a fair few announcements around the future of electric vehicles (EVs).
I’ve been playing with the numbers and creating some simple quantitative models: part of my work to develop a new training course looking at the impact of EVs on power systems. Below I’d like to share some early insights based on these calculations, along with a simple online tool so you can play with your own numbers too.
Electric cars… what about the electricity?
Here in the car-loving UK, the government recently announced that if you want a new car after 2040, it will have to be electric. The UK isn’t the only country to have announced a timetable for the transition from fossil-fuelled to electric vehicles, at least at the light-vehicle/passenger car level. On a more local level, some city authorities have been even more aggressive in their efforts to remove oil-burners – diesel particularly – from their streets, to improve air quality. This common direction of policy has seen a shift from manufacturers too, recognising that they need to start shifting their product portfolio away from internal combustion engine (ICE).
For those planning future electricity systems and infrastructure requirements, the shift from ICE to EV brings up two key questions:
1. How will it affect power demands at different times?
2. How will it affect total energy demand?
Sometimes the first question is phrased as: “what if everyone wants to charge their car at the same time?” It’s something that has certainly caused much head-scratching, not least amongst system operators such as the UK’s National Grid (see this pdf: their recent, and unusual, response to alarmist newspaper headlines). The short answer to this thorny question is that charging will have to be smart, so that the demand from EVs can be managed. If everyone wants to charge at the same time: tough, they can’t! The long answer involves aligning various tricky and interconnected policy, pricing, consumer behaviour, smart infrastructure and energy storage issues.
I plan to come back to smart charging and power demand in a future article.
For now, let’s address the second question. Assuming we can manage its time-of-delivery smartly, how much more total energy will we need to produce, to keep all these EVs running?
There have been plenty of detailed studies on this, undertaken by greater-resourced organisations than mine, so I’m certainly not aiming to repeat them. Nevertheless, whenever I investigate a topic I find it insightful to play with some of the fundamental variables around it. That way, if I’m reading future forecasts and analyses published elsewhere, or presenting their findings in the course of my training work, I’m better placed to discuss why (and by how much) these findings might prove inaccurate when the actual future arrives. The aim of this article is to share a brief flavour of that insight.
Here in the UK, one in-depth study is National Grid’s 2017 “Future Energy Scenarios” (FES), which includes specific modelling of the growth of EVs. You can find various documents around FES, including a detailed spreadsheet with their forecasted numbers.
In their two most aggressive EV scenarios, one has EVs as two-thirds of cars on the road in 2050, another 100%. They predict an added electricity demand of 35-45 TWh per year as a result of these EVs, constituting 9-11% of total electricity demand at that time.
In percentage terms, are those energy increases smaller numbers than you expected? They were to me.
They are small but certainly not insignificant.
To put it in power plant terms, you could get 40TWh a year from 5GW of new nuclear power plants (operating at 92% capacity factor) or from 10GW of offshore wind (operating at 46% capacity factor). With rooftop solar here in the UK operating at more like 12% capacity factor, you’d need another 38GW of solar. Alternatively, you could operate today’s existing 35GW of gas capacity at 54% annual capacity factor rather than its current 40%.
In practice, it would be a mixture of all these (and other) approaches.
Key electricity demand variables
Rather than try to recreate a complex forecasting model like that used by National Grid, I decided to take some basic variables, then run simple calculations to see how sensitive future EV electricity demand will be to them. I’m not interested in arriving at a “right” answer, or in suggesting that other forecasts may be wrong. I do want to better understand where and by how much changes in different input variables can change the conclusions drawn.
I included a simple calculation of the in-use CO2 emissions of EVs compared to their ICE cousins too. That’s to address the question as to whether we are simply shifting tailpipe carbon emissions to power plant carbon emissions.
In my sensitivity analysis, I’ve considered these key variables:
- The number of cars on the road
- The annual distance they travel
- The efficiencies of both EV and conventional (ICE) cars (battery energy or fuel volume used per km)
- The carbon intensity of electricity generation (carbon dioxide emitted for each unit of electricity generated)
Also of relevance are:
- The total underlying electricity demand (without EVs)
- Grid or “T&D” losses: how much electricity is lost getting from power plants to the EVs. The requirement for new electricity generation must exceed the end consumption by EVs to allow for this.
For conventional vehicles, the carbon dioxide emissions from burning a volume of fuel is another input. That’s not a “variable” in my calculations (which assumes conventional cars are burning oil-based petrol or diesel). However it could be if you wanted to compare emissions between EVs and alternative fuels: biofuels or hydrogen, for example.
I’m most interested in these outputs:
- The total additional electricity demand created by EVs
- What percentage increase this represents over underlying (other) electricity demand
- How much CO2 an EV puts out each year (via power plants meeting the increased energy demand), compared to how much CO2 a conventional car puts out each year (from its exhaust pipe).
For emissions, I’m focusing on carbon dioxide and on emissions “in-use” (i.e. caused by driving the cars around).
I’ve taken it as unarguable that for other emissions (NOx, particulates and so on), spitting these out locally from an exhaust pipe close to where people are living and walking creates worse health outcomes than anything emitted from a distant power plant. On the other hand, I haven’t considered other pollution and sustainability issues; for example associated with the manufacturing and end-of-life recycling of different car types. The raw materials for batteries is certainly a topic which worries some. Those are complex issues for another time!
It’s always a good start to check that any figures I produce with my very simple calculations are in the same ballpark as those calculated by more comprehensive models.
National Grid’s “two degrees” scenario (their “greenest”) assumes that by 2050 there will be 25 million cars on UK roads, all of them EVs. In their scenario, total electricity demand in 2050 is 392,000GWh, including T&D losses (6%). Of this, EVs account for 35,000GWh. Electricity generation is all zero carbon.
National Grid state that their “two degrees” model includes assumptions such as: future vehicles will have “lower energy demand” and “50% of vehicles will be autonomous” with sharing of these cars “reducing the distance traveled per person” but “increasing the mileage per vehicle”. As a reference point, the current average distance travelled per car in the UK is 12,500km. It’s important to note that this represents a downward trend from 14,700km fifteen years ago.
The current official consumption figures for a Nissan Leaf are 150Wh per km. If I assume that by 2050 that’s down to 100Wh per km (“lower energy demand vehicles”), then my simple calculator produces the National Grid’s 35,000GWh of new EV electricity demand by assuming each car travels 13,150km each year.
I’m not aware of a simple, average “km per year” number in National Grid’s published documents. Nevertheless, I’m comfortable that my outcome is consistent with today’s trend of decreasing annual usage per car being offset and reversed by National Grid’s assumption that half of future EVs are used for journey sharing; and hence travel further. [For example, if average annual distances for non-autonomous vehicles have dropped to 10,520km by 2050 but autonomous vehicles travel 50% further, then that produces my overall average of 13,150km per car per year].
Fig 1. Summary results table, “two degrees” highlighted (see other scenarios later on)
Notes: The “Total annual CO2 saved” calculation compares total annual emissions from the same fleet size and usage, one 100% EV and the other 100% conventional (using the car efficiency and power generation emission inputs shown). For calculating the equivalent power generation capacities to meet new generation demand, ‘CF’ refers to assumed capacity factor.
So, satisfied that my simple calculations are “in the right ballpark”, let’s play around with some different assumptions.
“Business as usual”
There are currently 31.1 million cars in the UK, on average driving 12,500km per year. (You can find those and other interesting mobility stats here).
Although the official consumption figures for a Nissan leaf are 150Wh per km, I’m a sceptic when it comes to whether official figures are achieved in real driving! I’ll assume instead that real electricity consumption is more like 200Wh per km.
Assuming no change in either the number or usage of cars, and based on my “pessimistic” electricity consumption per EV, switching all cars over to EVs would require 83,602GWh of new electricity generation. That’s an increase of 25% on current generation. It’s more than twice that required in National Grid’s “two degrees”, green scenario. It requires over 10GW of new nuclear or 20GW of offshore wind, for example.
Fig 2. Summary results table, “business as usual” highlighted:
What about the impact on carbon emissions?
The UK currently emits around 300g of CO2 per kWh of generated electricity. Assuming this too doesn’t change, each of our new EVs will be responsible for 806kg of CO2 emissions per year. A fleet of petrol cars of the same size and used in the same way would need average fuel economy of 170 km per gallon to match this (for UK readers: that’s 106 miles per gallon). At present that sounds outlandish – but be aware that manufacturers such as Mazda claim figures “well over 100mpg” will be possible from petrol engines in just a few years’ time.
Another way of looking at the emissions issue is to ask: how high would electricity generation emissions need to be, for EVs to be as “bad” as ICE?
Let’s say the latter do achieve a real-world economy of 100mpg (160km per gallon). Emissions from electricity generation would only need to rise to 320 gCO2 per kWh for ICE and EV to be at parity, in terms of annual emissions per vehicle.
That’s not much!
It’s hard to see why the UK would go backwards on electricity emissions. However in countries burning significant coal in their mix, emissions from EVs will be much higher than here in the UK. As a reference point, 100% coal would mean a “carbon intensity” of 900-1000 gCO2 per kWh.
Let’s say the UK’s generation emissions were 500 gCO2 per kWh. Now an ICE efficiency of 102km per gallon (just 64 mpg) would be “carbon competitive” with EVs.
So the point at which EVs become lower carbon than advanced ICE vehicles will depend on how electricity is generated – as well as how advanced ICEs can become. That’s an obvious point, but it’s nice to put some numbers on where the crossover might be. The chart below considers this balance in more detail: combine improved EV efficiency with cleaner power generation, and developments in ICE technology will have to be very dramatic in order to remain carbon competitive. (Reminder: this analysis ignores differences in carbon intensity during manufacturing and end-of-life).
Fig. 3 Carbon equivalence curves for different power generation emissions
The chart plots the efficiency of EVs on the x-axis (increasing to the right, less kWh needed per km) against the efficiency of conventional cars (increasing upwards, more km achieved per gallon of fuel). The lines join points where, for the indicated power generation emissions (g of CO2 per kWh generated), the annual emissions from an EV matches that from a conventional car. The grey bar indicates the current range of EV efficiency. As EV efficiency improves and/or power generation emissions fall, conventional fuel efficiency needs to increase, hitting the curve in order to be equally “clean” (at least in terms of carbon emissions during use)
My “business as usual” scenario is certainly pessimistic.
Even assuming no change in the number of cars or the distances they travel, it is surely certain that the efficiency of EVs will improve. Even if they only achieve today’s official consumption figures of around 150Wh per km, that immediately drops their electricity requirement from 83,602GWh per year down to 62,702GWh.
That’s still a 19% increase on current generation – but, compared to “business as usual”, it’s over 2.5GW of nuclear no longer needed (or 5GW of offshore wind and so on).
In reality, UK emissions from power generation have been shrinking, as we’ve been removing coal from the mix. Coal is due to be gone completely within the next decade. With renewables set to continue their growth and new nuclear scheduled to be built, it seems reasonable to assume emissions will drop further. Let’s assume they drop from today’s 300 gCO2 per kWh down to 225.
With each EV now responsible for just 454 kg of CO2 per year, petrol cars would need to achieve real-world efficiencies of over 300km per gallon (around 190mpg) to be carbon competitive. I doubt any amount of new ICE technology will achieve that?
This reiterates the conclusion that was obvious from Fig. 3 above. Shifting to EVs, when done in parallel with continued cleaning of the electricity supply, is undoubtedly a lower-carbon option compared to continuing with ICE, even if the latter continue to make large efficiency improvements. In the long-term of course, many hope that the electricity supply should be zero-carbon.
A key assumption here is that the UK can add almost 63,000GWh more electricity to its mix at the same time as cutting out carbon-emitting generation from this mix. That might not be the case if a significant chunk of this new demand was met by burning natural gas, for example. A thorough discussion of that would need a separate article: after all, there are many ways to make a theoretical electricity mix add up. It’s fair to say that many – most? – analyses (National Grid’s included) suggest nuclear will remain important in any future, UK low-carbon mix.
Fig 4. Summary results table, “continued decarbonisation” highlighted:
It’s been a global trend for some time that populations are being concentrated into urban areas. So in future more people will live, work and play in cities. Public mass transport makes much more sense as a way to get around in those environments: on central London roads, average speeds as low as 8 miles per hour have been recorded.
I’ve read different opinions on car buying trends, but some suggest that younger generations are either less interested in owning cars or that, even when they do, they use them less. Car sharing, taxi/Uber services and – eventually – autonomous vehicles could all have a huge impact on how many cars exist and how far they go.
What if, in future, the UK has a third fewer cars – say 20,000,000 – and, even when shared, they travel no further than today? (A more disruptive version of National Grid’s “two degrees”, if you like). Unlikely? Probably, but then it’s always good to test extreme scenarios.
A new world record for EV efficiency was set this year at 88Wh per km in a Tesla. If that’s what’s possible now, then who’s to say that consumption as low as 80Wh per km won’t be possible in future? Technology will have moved on, new materials will have made cars lighter and autonomous vehicles will be programmed to operate much more efficiently than impatient human drivers do.
What if much of the energy used to charge EVs is generated locally, halving overall grid losses to around 3%?
Put all these assumptions together and that whole EV fleet can be kept running with an additional 20,620GWh of electricity generation; just 6% more than today’s total. Just one new nuclear plant or four or five large offshore wind farms would cover that; or – given that I talked about local generation – perhaps 20GW of new distributed solar capacity.
Fig 5. Summary results table, “transport transformed” highlighted:
What if underlying electricity demand changes during the time it takes to make our transport transition? Demand might fall if efficiency gains continue. It might rise if we start to electrify heating, or if the country sees a post-Brexit manufacturing boom (feel free to challenge one of my assumptions there). Both government and National Grid forecasts are for electricity demand to rise.
If underlying demand is 25% bigger than today, at around 411,000GWh per year, that means we’ll have needed to increase supply by 82,250GWh anyway, even if we do nothing to electrify transport. In other words, the latter (changing to EVs) is a problem a quarter the size of the former (growth in other demand). Conclusion: don’t forget to think about other changes in demand when thinking about the impact of EVs on the electricity system as a whole. Maybe we’ll be building or replacing new infrastructure anyway?
Certainly I’d suggest that demand reduction and efficiency needs adding to clean generation as a key policy goal to continue in parallel with EV growth. With it, the transition to EVs will be much easier (and less expensive) than without.
Explore the numbers yourself
Rather than read about me playing with more scenarios, why not try some for yourself?
To help you, I’ve turned my simple calculator into an online tool which you can find here.
Go ahead, challenge the assumptions that I and others make. Tailor the inputs for your market (taking care to use the right units). As you change the inputs, focus on thinking about why they might change in that direction or by that amount: does it depend on technology, policy, behavioural trends, economics? Or, most likely, on all these and more?
I can’t promise you’ll predict the future as accurately as those who spend months developing complex models and scenarios, but it should provide a useful illustration of how the different variables interact.
More than that, it should provide a trigger to think about the issues and assumptions involved in setting these variables: all of which need to be questioned. After all, this is ultimately a question of action “on the ground”, not numbers on a spreadsheet. Which actions are realistic and which are not? How does this change the numbers?
Finally, don’t forget that this analysis is only concerned with making the bulk energy numbers add up and the decarbonisation argument make sense. It doesn’t say anything about delivering this energy to the right place at the right time; and hence how this impacts power and grid capacity requirements. Those are issues for another day!
(Interested in exploring these issues further? The training course Electric Vehicles & the Power System: Impacts & Integration will be arriving in 2018. Email firstname.lastname@example.org with the subject line “EV Course” if you want to be kept informed about it)