We recently ran a perspective piece on Tesla and electric vehicles in which John Petersen suggests that "a grid-powered electric vehicle might make individual drivers feel warm and fuzzy about themselves, but from a public policy and resource-conservation perspective, it’s the most wasteful plan ever devised." We had a rebuttal from NRDC attorney Max Baumhefner. Nick Butcher took a close look at whether there is a battery material resource limitation to widespread electric vehicle penetration in Part One of his EV investigation. In Part Two, he looks at whether EVs are really ‘clean and green,’ or are just misrepresented as such.


Tesla’s Model S has been on the road, with private customers behind the wheel, for almost a month. The styling and performance will certainly draw attention, and the early reviews are nothing short of glowing, but do these new owners really deserve to be viewed as enablers of a green revolution?

In the first article of this series, "EV Myths & Realities, Part 1: The Battery Crisis," I looked at the supply and production constraints on battery availability for EVs and demonstrated that they could readily scale to meet any foreseeable demand. In Part 2, I will discuss whether EVs are really ‘clean and green,’ or are just misrepresented as such.

This is an important issue for investors and buyers alike. EVs are seen as a big step forward on the path to a clean and energy independent future -- but if they only move the pollution over the hill to a coal power plant, then their market position and consumer adoption may suffer, spelling bad news for the EV trailblazers. Whether you’re looking at taking a long position in an EV manufacturer or buying an EV as an informed consumer, you should be confident that the products’ 'green and sustainable' credentials will last the distance.

Let’s start with three key objectives in mind:

  • Objective 1: Reduce CO2 emissions dramatically. If you don’t think this is important, then, well, you’re probably not a climatologist. More importantly, you’re certainly not going to care about the aspects of this article that relate to relative CO2 emissions. For everyone else (and particularly prospective EV buyers), I think we can take this goal as given.
  • Objective 2: Improve energy security, preferably at a national level, but most critically at a global level. If you don’t have a particular resource but someone else does, then you can generally trade with them -- wheels of commerce and all that. If neither of you have the resource, however, and you both want it, then things are going to get tense. Energy security means there's plenty to go around.
  • Objective 3: Be sustainable. The future is hard to predict, but we can at least avoid things which we are near certain will bring us grief in 50 years.

Addressing these objectives in enough detail to reach a conclusion is going to take a fair bit of space, and even then I’ll rely on the vast amount of existing work in the field for the details. To try and keep things under control and in readily digestible portions, I will split this post into four sections: Greening the Grid, Reaching Break-Even, Ramp-Up and the Integral Effect, and Considering Complexities.
As part of keeping things under control, I take a fairly U.S.-centric approach in this analysis. This is not an inherent cultural bias;  rather, it's just that the issue is hard enough to address looking only at one country, let alone hundreds. As a large, diversified, and highly developed economy, the U.S. is a good representative of the challenges faced by the world as a whole when it comes to clean mobility. If anything, our challenges are greater, as the U.S. has generally been somewhat slow (historically, though this is recently improving) to adopt emissions reduction initiatives in the electricity sector, and has a huge vehicle population and associate infrastructural lock-in.

If we can demonstrate that problems can be solved here, then we’ll have a template for solving them elsewhere (as opposed to, for example, solving the emissions conundrum for Switzerland, France, Norway, etc., which already have extremely ‘clean’ electricity and hence represent a somewhat trivial example). The other great thing about the U.S. is that it’s the subject of a lot of highly detailed and comprehensive research and studies, all published in English, which saved me a lot of time.

Greening the Grid

Starting with a bird’s-eye perspective, this whole topic is basically a question of different energy sources: how clean they are, and how abundant.

Oil from fossil sources is neither clean (11kg+ CO2 per gallon burnt, to say nothing of accidental spills) nor secure, with grave concerns about the rate at which additional production capacity might be bought on-line to meet new demand -- concern that remains even considering unconventional sources. Electricity can be very clean, but it can also be very dirty -- as is the case with old coal power plants. Supplies are more secure, however, ranging from several decades' worth with natural gas, to more than 100 years with coal, to millennia with advanced nuclear, to eons for solar and wind.

Broadly speaking, then, Objectives 2 and 3 are taken care of: the diversity of sources available to generate electricity, including many highly secure and sustainable on an order of millennia, mean that while there is a strong likelihood of short-term price volatility and periodic shortages, the mid-term "electricity supply" picture is generally positive. The electrical vehicle energy supply challenge is basically one of emissions. Unfortunately for those who were hoping for a short article, addressing that first objective is going to be much more complicated.

The trend, as economies become more advanced, is universally one of increasing electrification, and creating that electricity in a "clean" manner is one of the biggest challenges we face -- to the extent that Bill Gates (no slouch on technology or policy) said he’d dedicate his "one wish" to finding a solution. (To his credit, he’s not just wishing.)

To put the problem in perspective, in 2008 the U.S. Transport Sector emitted 1.93 gigatons of CO2. The same year, the U.S. Electricity Sector emitted 2.4 gigatons, with coal responsible for 82 percent (despite delivering only 45 percent of the total energy). Since both secure electricity and low CO2 emissions are both critically important to our future quality of life, basically every major public policy entity acknowledges the need to massively decarbonize the electricity sector. The International Energy Agency says complete decarbonization by the second half of this century is the minimum we should aim for. Many other entities advocate even more aggressive reduction targets to limit the extent and impact of global warming.

Thankfully, we have a pretty broad base of technologies to draw on to achieve massive emissions reductions in the electricity sector, even while increasing total generation. In the U.S., the average is around 600 grams of CO2 per kilowatt-hour, but the only modern generation technology worse than that is coal.

The above chart shows the emissions in grams of CO2 per kilowatt-hour of the majority of existing generation technologies. There’s no true consensus on this -- I’ve taken the data from a very good IEA study that aggregates the results of numerous other studies, and added in the exact reported values for a number of newer products (USC, A-USC, & CCS Coal, Gas CT, Gas CC). Generally the green and purple numbers above represent "new" power plant options, with the red numbers being outdated solutions or unusual circumstances. In the developed world, coal is actually in decline, caught between the dual pincers of both public opinion and economics. A coal phaseout is strongly advocated by many environmental agencies, and the telltale signs are clearly evident in planned capacity increases in the U.S. Conventional coal is dirty, CO2-intensive, comparatively inflexible, and no longer competitive for new build.

We have the makings of the solutions we need, and we clearly have a strong incentive to make the shift with near-universal consensus that "clean, renewable electricity" is the way to go. The challenge is going to be making the switch in a sufficiently short time. If we do, we win -- if we don’t, then any discussion of transport sector emissions reduction is a little irrelevant anyway.

Reaching Break-Even

The real question is, "Are EVs cleaner than internal combustion engines (ICE)?" Answer: It depends on the emissions profile of the electricity you put in. Garbage in, garbage out.

Do we need to get the grid to carbon-zero before EVs become cleaner than ICE?

No, and not by a long shot. There’s a number hiding here that represents the electricity generation CO2 per kilowatt-hour at which electric vehicles becomes cleaner than ICE equivalents for a given combination of:

    • MPG – Miles per gallon for the ICE vehicle
    • kgCO2/Gallon – 11.2 (approximately) for conventional oil
    • Wh/mile – Watt-hours of electricity per mile for the EV, include line (6 percent) and charging (10 percent) losses
We also need to include another number: gCO2/Mile embodied, which captures the difference between the embodied emissions of the EV and the ICE. I take this as 3 tons CO2 to the EVs debit, giving a 17 gram per mile advantage to the ICE (12,000 miles per annum, 15-year life).

The figure below shows the relationship between these numbers, for all combinations of EV wh/mi and ICE MPG.

Shown are ranges from 10 to 70 mpg and 250 to 500 wh/mi, with the ‘break-even’ generation gCO2/kWh calculated and plotted for each point. I’ve thresholded the ‘break-even’ generation at 1000g/kWh, as by the time we reach that level of results, it’s pretty clear the EV is preferable and it gives us better resolution at the lower numbers we’re interested in. It’s not even close between an efficient EV (say, 300 wh/mi) and an average car (say, 30 mpg) -- the EV is cleaner even when running off a filthy coal plant. In the foreground, however, we see a challenge: an inefficient EV (500 wh/mi) would need to be charged with electricity that averaged less than 300 g/kWh in order to beat an ultra-efficient 70 mpg ICE car.

The purpose of the last chart is mostly to give you an idea of the overall landscape (and to add some much-needed color). This next is a contour plot of the same data, much more useful as it allows us to read comparisons directly.

You’ll note I’ve added several straight lines with the names of particular vehicles. (For the EV wh/mi numbers, I’ve used the EPA figures, which include charging losses, to which I’ve added 6 percent to account for transmission loss.) The contours map onto the color chart on the right; each contour is labeled with the gCO2/kWh it equates to.

Now we can do quick comparisons -- life is good.

    • Nissan Leaf vs. Prius: Leaf wins for gCO2/kWh < 580
    • Honda (HMC) Fit vs. VW Polo BlueMotion: Fit wins for gCO2/kWh < 520
    • Tesla S vs. Lexus GS: Tesla wins for gCO2/kWh < 870
The only thing not considered in this chart is the increase in CO2/gallon for gasoline as extraction becomes more difficult. A Prius running on gasoline from a Coal-to-Liquids plant would have an MPG equivalent of only 25 on this chart, meaning it would be worse than a Tesla S running off a coal plant.

Combining what’s shown here with what was shown in the previous section regarding the need to green the grid, we can happily conclude that even if we can only charge our EVs with energy from a modern gas power plant, they will equal or better the emissions of the most efficient ICEs. The Tesla S beats its most efficient segment competitors even when charged with decidedly average coal power.

Ramp-Up and the Integral Effect

Now that we can see the relationship between generation emissions and EV "cleanliness," you could be forgiven for assuming that we shouldn’t begin switching to EVs until we can be sure that the grid's marginal emissions are below the critical point. After all, why introduce a new vehicle technology only to connect it to a coal power plant with net emissions even worse than oil? We know the grid needs to "green" -- should we not wait for that to happen?   

The easiest statistic to use for "grid greenness" is the average gCO2/kWh for the region in question, but this gives arguably flawed results. To establish the worst-case consequence of adding EVs to the grid, we should consider not the average but rather the marginal emissions. Marginal emissions are best understood as the emissions profile of the generation asset that’s generating electricity because you are charging your car -- and would not be otherwise. This is actually an extremely complex problem to solve (we must consider the generation mix for the region, what’s running anyway, what the transmission constraints are, etc.), and we will look at those factors in the "Complexities" section. For now, let’s just consider the ‘average of the worst’ that are likely to be on the grid -- some mix of coal and gas, trending to gas as coal is phased out, and trending eventually toward the ‘clean options’ as we close in on that IEA goal toward the middle of the century. Remember it’s the average marginal emissions that matter; if your car is charged, considering the yearly average, on 20 percent nuclear, 20 percent Gas CC, 20 percent Gas SC, 20 percent coal, and 20 percent hydro, then your average marginal emissions will be around 380g/kWh.

The profile I have assumes a far worse mix than this (extremely pessimistically so, as we’ll see later), with the ‘marginal EV’ emissions starting at 900gCO2/kWh, decreasing slightly below the existing U.S. average by 2020, reaching 220g/kWh in 2050 (lagging well behind the IEA target), and finally closing in on ‘clean’ around 2100. Really, we should aim to be a lot faster than this to avoid the worst of climate change, but let’s wear our negative hats. It could be argued that I’m being unreasonable in the speed of my forecast early reductions, were it not for the average marginal emissions already being well below my starting figure.

The reason the reductions start quickly and then get slower is that the cleaner the grid gets, the harder it is to improve. Shutting down old coal and replacing it with clean coal, gas, etc. is quite easy. Replacing clean gas with CCS, solar, nuclear, wind, etc. is harder (though, ironically, wind and solar are much easier to manage with a grid full of EVs). So now we have an, admittedly pessimistic, estimate for when the Nissan Leaf becomes ‘cleaner’ than the Prius: 2022. Should we wait until then to begin introducing EVs, or at least subsidizing their introduction?

The answer is no: the time to begin introducing EVs is now, and the reason (aside from the fact that many grids are already sufficiently green at the margin, as we’ll see in Section 4) is found when we look at new technology adoption rates, market share, fleet penetration and replacement intervals.

Shifts between technologies usually obey a logistic curve. For EVs, this will be evident in their market share as a fraction of total vehicles on the road. I assume the following profile of new vehicle sales in the U.S.: EVs increase to 50 percent of annual sales in 2033, saturating at 90 percent in 2053 (100 percent would be better, but some ICE may well remain, even if they run on synthesized zero emissions fuel).

Share of the total vehicle fleet will lag this ratio considerably though, using the generally accepted 15-year time between vehicle purchase and retirement. The chart below shows the EV fraction of the total U.S. vehicle fleet, first for the case shown above, and second for the case where we delay introduction by ten years (waiting until the Leaf is certainly cleaner than the Prius under our speculative marginal emissions profile).

The area between those EV early and late curves is the vehicle-years of additional ICE emissions that result from waiting, though of course it’s also vehicle-years of EV emissions that are avoided.

We’ve compared early EV adoption with late EV adoption in terms of the time taken for EVs to penetrate the vehicle fleet, but what we actually care about is the effect on the cumulative CO2 emissions. (Note: cumulative CO2 -- the CO2 humanity releases into the atmosphere in a single year -- wouldn’t, in isolation, have a huge effect. The problem is that we do it year after year, and in ever-increasing quantities.)

Above, I've presented all the data we need as an input to calculate the cumulative CO2 emissions from the U.S. vehicle fleet under our scenarios. Now it's time for a look at the output.

Here we see that, whatever happens, the U.S. light vehicle sector is still going to make a huge mess over the next century. Behavioral changes to reduce that are really welcome. But the point here is the relative emissions of our various EV adoption scenarios. You’ll see I’ve thrown a third scenario in the mix, where we never shift to EVs but rather just see rapidly increasing ICE efficiency.

The best outcome of the three is an early switch to EVs, at 46 GT of cumulative emissions. The small penalty we pay in the early days (running them on the dirtier grid) is completely insignificant (for those counting, the Early scenario results in cumulative emissions in 2016 of 7.091862 GT, while the late scenario results in 7.091377 GT) and more than compensated for by the benefits that come from more rapid fleet penetration. Waiting until the grid is ‘green-ish’ results in an increase of 3 GT of cumulative emissions; abandoning EVs altogether in favor of high efficiency ICE means an increase of 23.2GT (and ongoing beyond 2100).

You might be a little shocked that, even if we start moving to EVs now, the light vehicle sector will still emit 46 GT of CO2 by 2100. That’s the penalty of slow replacement, and the fact that in my model almost 20 percent of the vehicle fleet is still ICE at the end of the century. EVs themselves are responsible for only 7.3 GT of emissions in that time, and at the end of the century, despite accounting for 80 percent of the vehicle fleet, EVs will have annual emissions of only 0.04 GT running on our clean grid.


Even with fairly dirty electricity, EVs are cleaner than ICEs. We only need to get to an average marginal generation CO2 intensity of 500g/kWh to have a Model S be cleaner than a Prius -- and many markets are already there. The only modern generation technology that is worse than this is coal, and even it can be improved with carbon capture & storage technologies. Comparing like-for-like, the Model S Performance is cleaner than a Mercedes E400 hybrid even if it’s running on a reasonably modern coal plant output.

If you see a new EV buyer lobbying for new coal plants, then you can perhaps accuse them of environmental ignorance. Otherwise, give them a pat on the back: they’ve put a decent chunk of their money on the line and braved a new technology to make the switch.  As long as we don’t totally fail the energy generation challenge, it’s a switch with a huge net benefit for everyone. EVs are not perfect, but they beat the snot out of any like-for-like midterm alternative on the radar today.

Nick Butcher is an engineer from New Zealand, now living in Switzerland. He has almost a decade of experience in product development in the energy industry, most of it relating to grid interactive power electronics applied to a range of applications including electric vehicle fast-charging and grid battery energy storage, both with engineering multinational ABB. He now splits his time between Ampard (an energy storage startup), SwissKitePower (Kite Energy), several other embryonic ideas, and freelance work.

Disclosure: The author has a long position in TSLA.