0. Some Terminology, Ground Rules and Basic Reality Checks

EV – this name will be use as an umbrella term for any vehicle that can drive based on power charged from a plug. This includes the “pure” electric-only BEV (battery EV) like the Nissan Leaf and the Tesla S, the PHEV or plug-in hybrid and plug-in Toyota Prius, and also more exotic beasts like the EREV (extended-range EV, e.g. the Chevy Volt, whose gas motor can only be used as a generator to recharge its battery, but not to directly drive the car – h/t Morgan in Austin for the correction regarding the Volt!).

ICE – internal combustion engine.

GHG – greenhouse gases. In practical terms this is mainly CO2, with sometimes-substantial contribution from methane. The GHG footprint is usually expressed in terms of kg-CO2-equivalent.

LCA – life-cycle analysis. Nowadays, this refers to any attempt to quantify the real-life GHG footprint of some activity of product, from its inception to its end of life.

Other acronyms will be defined at the time of their introduction. Finally (for this section), the reality checks.

Reality Check 0-A: GHG footprint analysis is a young and still-evolving science.

Wide awareness of human-caused global warming, even in the academic community, did not really take place before the 1990s. Therefore, while other types of pollution have been studied for decades, GHG footprint analysis is still an emerging field. So no footprint report you read,anywhere, should be considered as the final definitive word. Rather, all are attempts (some of higher quality, some of lower) to get some general stab of the GHG impact of various activities and products.

Reality Check 0-B: All major GHG-footprint LCAs of EVs are done with good faith.

I’ve read at least a couple of EV LCAs that made me mad. They were unfair; they were sloppy; etc. etc. But I do not concur with the various accusations against the authors of such reports. No one who engages in GHG-footprint LCA does it to make the environment worse. Some authors probably do have a deep-seated belief that EVs are not good, and it might affect their analysis choices and surely affects their report’s framing – but they are not shills for anyone, least of all the oil industry.And even if there was such a shill to be ratted out, for our own sanity and well-being it’s better to critique their work on the merits. Sort of in line with the community culture we’re trying to promote here, right?

Reality Check 0-C: EVs have been subject to far more GHG-footprint scrutiny than ICE vehicles.

This is the height of irony; after all, EVs are still a fraction of a percent of the global vehicle fleet. Surely we’d like to know better about the impact of the 99.9x%, before we slice and dice the remainder? But this is a natural result of GHG-footprint’s young history: the science was hardly born, when this slew of new and intriguing vehicle types came up as possible mitigators of ICE problems (not just EVs, but also stuff like liquified-natural-gas and hydrogen-cell cars). Naturally it is more exciting, and at face value more responsible, to inspect the GHG footprint of these novel products.It seems more boring to go back and scrutinize petroleum the the vehicles guzzling its products for the umpteenth time, right? Except there is no umpteenth time, there is hardly a first time here. How to I know? I asked Dr. Andy Burnham of Argonne National Labs, responsible for one the most-relied-upon suite of vehicle-related GHG analysis. He is the lead author of the most recent article out of ANL to mention gasoline and ICE-vehicle GHG overhead. His article focuses on natural gas; data for gasoline relies on a decade-old analysis. No one in their lab has seriously touched the subject since the early 2000s. And just like with Internet history, in GHG-analysis history the early 2000s are essentially the dinosaur era.

So just keep this in mind: it is a verifiable fact that EVs are being singled out in the present discourse about vehicular GHG-footprint analysis. This is not out of bad faith, but simple group dynamics, and the notion that ICE-vehicle footprint is a known and settled quantity. btw, I beg to differ with this notion. One small example: I haven’t seen LCAs that seriously address the gradually increasing amount of maintenance and repair that ICE cars require as the car ages. A larger example will be discussed towards the end.

Reality Check 0-D: Even the most EV-skeptic GHG-footprint LCAs estimate that EVs are better than typical ICE compacts, and equivalent to ICE hybrids.

You wouldn’t guess it from the vibes in certain sectors of the environmental movement. But even those LCAs least favorable to EVs (herehere), still place it clearly ahead of most ICE cars – surely the gas-guzzling US fleet where even new compacts barely average 25 MPG – and roughly on par with ICE hybrids. The more EV-friendly analyses, which include this Union of Concerned Scientists report, as well as the most recent official word from the EPA itself, have the EVs beat hybrids fairly handily. Keep this in mind…On to the actual analysis. LCA for vehicles is best (IMHO) decomposed into 3 concentric circles of analysis:

1. Energy efficiency – how much energy it takes to drive a mile?
2. Ongoing-use footprint – suppose there’s an EV and an ICE car (or ICE hybrid) in the parking lot, and you can take either one. How much GHG emission is associated with driving a mile in each car, including the overhead from producing the elecricity on one hand, and taking the crude from the ground and turning it into gas at the pump on the other hand?
3. Life-Cycle Analysis – the overall amount of GHG emissions generated by the car, including #2 for all the miles driven on it, and also production, maintenance and decommission overhead for the car itself.

1. Energy Efficiency and MPGe

I am a great fan of MPGe. It is arguably the cleanest measure in this entire vehicle-footprint analysis mess, and it brings us back to the basics.

MPGe stands for Miles-Per-Gallon-equivalent. The term was invented by the EPA to enable energy-efficiency comparisons between ICE and non-ICE cars. It estimates the average distance a non-ICE car can travel, while using an amount of energy exactly equal to one gallon of gas.

EPA MPGe and range sticker of the 2013 Nissan Leaf
EPA’s energy efficiency sticker for the 2013 Nissan Leaf

Why do I like MPGe? I mean, besides the fact that EVs fare fabulously on it? Well, as we are often reminded, the CO2 and global-warming crisis is the most extreme manifestation of a broader challenge facing our society: can we sustain acceptable quality of life while addressing the urgent imperative to use far less energy per capita than we currently do?

It is therefore refreshing to learn that you can have a perfectly good car fulfilling most or all of your family-locomotion needs, and still use so much less energy. Even the less-efficient members of the leading EV family (that would be the Volt and the Model S) are 2x as efficient as the Prius and 3x-5x more efficient than regular ICE cars. The Nissan Leaf and other compact BEVs are about 2.5x more energy efficient than the Prius (115-120+ MPGe, vs. 47 MPG). And if you take a compact BEV to replace the in-city miles of an ICE car, then the BEV advantage is 4x-8x, because the city is precisely where the EV shines while non-hybrid ICE vehicles struggle. As a case in point, most of the 4300 miles we’ve driven so far on our 2012 Leaf (106 MPGe EPA in-city, and according to the car’s log we’ve done somewhat better than that) were in-city, where our 2001 Santa Fe does at most 15 MPG (my estimate based on actual gas consumption). That’s >7x improvement right there.

Reality Check 1: In terms of pure energy efficiency, the worst of EVs are about 2x better than the best of ICE hybrids. In comparison with non-hybrid ICE the gap is larger, even more so for urban driving.

You can easily find online polemics trashing the MPGe. These attacks are the result of a misunderstanding: they mix step #1 and step #2 of the footprint analysis. Of course, step #2 is important – but the question of energy efficiency stands on its own, and as the one with the most precise answer, it is the best place to start.And now, to step #2:

2. Ongoing-Use Footprint Analysis

A.k.a., “wells-to-wheels” or “everything but the car itself” analysis. This is when things start getting dicey. Let’s start with what is supposedly the easier side: ICE cars.

The baseline GHG intensity of burning oil is about 270 gCO2/KWh (note that we use KWh to make the comparison with EVs easier; usually the numbers are given as gCO2/MJoule but 1 KWh=3.6 MJoule). However, this does not include the “wells-to-pump” overhead due to extraction, transportation, refining, etc. The estimate for the average overhead is somewhere in the 25%-35% range, it varies widely by oil source (pdf) and IMHO it is generally on the low side – a point I will get to towards the diary’s end. However, a “safe” consensus take-home number is ~350 gCO2/KWh for “wells-to-wheels” ICE driving.

Now to the EV side. EVs generally charge from the grid, so their ongoing-use footprint is highly variable from region to region. For example, our Seattle City Light grid is 90% hydro, 4% wind, 4% nukes and very very little fossil. The CO2 footprint of renewables and nukes is puny: the estimates for practically all of them are well under 100 gCO2/KWh, usually far less than that. (aside: I happen to be a nuke-skeptic, and I haven’t scrutinized the nuke emissions calculations to see if they miss something big – but this is a bit moot at this point, b/c Fukushima has made nukes about as popular as The Plague).

How do we mesh all these numbers together? The ongoing-use footprint of any vehicle can be written as

{Source intensity (gCO2/KWh)}
Footprint (gCO2/mile) =    ————————————-    x   {33.7 KWh/gallon}
{Vehicle efficiency (MPG or MPGe)}

The KWh/gallon number doesn’t matter when comparing cars, b/c this number is just a constant conversion factor. The comparison is reduced to dividing the source intensity by the MPG or MPGe.For example: if we compare an EV running from a renewable/nuke dominated grid like Seattle’s with say 50 gCO2/KWh average footprint, to an ICE car, our 2x-8x advantage in vehicle efficiency is now multiplied by another 350/50 = 7x advantage on the power source GHG intensity – to an overall advantage of some 15x over ICE hybrids and way more than that over regular ICE. Seriously, I can probably drive 100 miles on our Leaf causing the emission equivalent of maybe 2 miles driving our Santa Fe in the city.

Seattle is not alone: All along the West Coast, hydro power dominates and wind is substantial, yielding a renewable-dominated grid. Incidentally, the West Coast is also home to 3 of the top 4 local EV markets, and roughly half of the EVs currently rolling on American roads. The Northeast, another area with above-average EV adoption rate, is dominated by nukes and hydro.

Reality Check 2a: in areas whose power grid is dominated by non-fossil energy sources – such as the West Coast and substantial parts of the Northeast – the ongoing-use GHG footprint of driving an EV mile is at least 15x better than driving an ICE hybrid mile, and as much as 50x better than driving a mile in a non-hybrid ICE vehicle.

“Unclean at any speed” is starting look rather doubtful at this point.On to fossils. According to the Intergovernmental Panel on Climate Change (IPCC), the median estimate of natural-gas power plant GHG footprint is 469 gCO2/KWh. This number includes the plant’s efficiency losses and other overhead contributions. It is a higher footprint than oil’s, but only by about 1.3x. Remember: EVs start out with a 2x-8x energy-efficiency advantage. Therefore, EVs drawing power from natural gas should be at least 20% better than comparable ICE hybrids in terms of footprint, on an ongoing-use basis. For example the Climate Central report lists Rhode Island’s grid as essentially all-natural-gas, and pegs Nissan Leaf’s ongoing-use GHG footprint there at 10% better than the Prius (they used a higher emission factor for natural gas than the IPCC figure quoted above; you start seeing the problem with all those snazzy footprint-analysis documents? The Devil’s in the details).

Reality Check 2b: in areas whose grid is powered by natural gas, the ongoing-use GHG footprint of driving an EV mile is about 1.3x-2x smaller than driving an ICE hybrid mile (depending on model), and 2x-6x smaller than driving a mile in a non-hybrid ICE vehicle.

US utility power source mix, state averages, 2012. From Climate Central's EV vs. hybrid report, based on Federal data.
US 2012 electric-utility fuel mix, averaged by state. Note the doubly poor representation of data: 1. Most utilities are not defined by state boundaries. The standard approach divides the US into 12 nationwide power-producing regions whose boundaries bisect many states, and 2. The figure distorts the vastly different power consumption of different states (e.g, Wyoming is shown in the same size as California). This sloppy and confusing data visualization is par for the course of the Climate Central report, from which the figure is taken.

Now please follow me on a small detour. Remember the caveat regarding the GHG footprint of driving a biodiesel car? In particular, the footprint of driving a vehicle on biodiesel generated from waste oil is 6x-8x smaller than driving it on “regular” diesel.

What gives? After all, waste-oil biodiesel also gets burned by the engine, with CO2 coming out the exhaust. Here’s what gives. The cooking oil is produced anyway for the food industry. And no matter how you dispose of it, the CO2 it traps will eventually be released to the atmosphere anyway. So the only GHG impact of waste-oil biodiesel is the energy needed to convert it.

Ok…  Now I’m ready to bite the black bullet: Coal.

Coal plants’ huge CO2 footprint is the reason some smartasses have invented the epithet “Coal Cars” to describe EVs. Truth be told, the US electricity grid in 2012 was only 37% coal, down from 45% in 2010; and as mentioned above most EV-friendly regions use far less coal than that.

But still. At an IPCC median of almost exactly 1000 gCO2/KWh, the listed footprint of coal offsets too much of the EV energy-efficiency advantage. At face value, a compact EV fed from a coal-dominated grid would now emit some 1.2x-1.4x more GHGs per mile driven, on an ongoing-use basis, than the best ICE hybrids. Leading analysts like the Climate Central authors to literally gloat.

Whoa. Hold your horses, gloaters. It takes coal plants at least several days to change their coal-burning rate. Which means that they burn their coal at the same rate, day – and night.What happens to all that CO2 burned up at night, when consumers use far less electricity? Some of it is converted to gravitational energy: every night utilities pump water up into water tanks and reservoirs. This enables gravity-based water distribution during the day, and also provides emergency water in case of blackouts.

But the overall available amount of water-tank energy storage is barely a few percent of power plant capacity (see, e.g., the data here). And there’s really no other massive storage option out there right now. The vast majority of off-peak burned coal goes to waste (waste heat, I suppose; although I could not get a positive verification for this).

Now… if only there was some additional, emerging form of energy storage that can be invoked at night… that would be a win-win for people and the planet…

…Are you kidding me? When do you think most EV drivers charge their cars? If they are substantial daily-commute drivers, they reach the evening needing to recharge most of their battery in time for the morning commute. Yes, in principle if they plug their car in right at 5 PM, their charging might coincide with the evening power-demand peak. But consider this:1. Recharging the EV battery on a daily basis from nearly-empty might take anywhere from a couple of hours to nearly all night, depending upon the charging method (I’m excluding DC fast-charging here, because that’s not a daily home-based charging method) and the charging amount needed.

2. All EVs come with a charging timer, so the fact you plug the car in at 5 PM doesn’t mean it starts swilling juice at 5 PM. You can set the timer to any starting time, as long as you are recharged by morning.

3. Coal-dominated utilities greatly reward off-peak charging. Examples: Georgia Power, supplying juice to Atlanta, one of the top 5 EV markets nationwide (and drawing ~40% of its power from coal) has these rates for EV drivers: 20.3c/hour 2-7 PM M-F, vs 1.3c/hour 11PM-7AM. Say you need to recharge 12KWh every night, your charger can do it in 2 hours, and you get home by 5PM. Would you rather pay nearly $2.5 every night… or set the timer to late-night and pay only 15c??.  In Lexington KY whose grid is nearly all coal, the difference is less dramatic, but still decent at 14c/KWh peak vs. 5.3c/KWh off-peak.

In short: EV drivers of the Coal Belt, fear not. If you diligently charge during the off-peak hours, your EV is not a “coal car” but a “wasted-coal-energy recycling car”, with a negligible ongoing-use GHG footprint. There are other utility types where off-peak charging matters, but the most dramatic effect is doubtlessly in coal regions.

Here’s a shocker: most current analyses of EV’s footprint don’t incorporate any assumption about EV charging patterns. Instead, they use the list value for coal’s energy intensity, which means that they assume EV charge patterns are the same as the daily patterns of overall consumer demand. A ridiculously wrong assumption.

Some people have started looking into this, just barely. An analysis mentioned in this book chapter from UC-Davis assumed 4 realistic EV-charging scenarios on a coal grid, and found about a 50% reduction in footprint compared to the formal peak-load footprint. I have a feeling the reality is far better than that, but I’ll take this number as an upper bound.

Reality Check 2c: in coal-dominated electric grids, if an EV charges mostly during peak-demand hours, then its ongoing-use GHG footprint can be as bad as 1.2x-1.4x higher than a comparable ICE hybrid.

However, realistic assumptions about EV off-peak charging reduce this footprint by at least one-half, and brings the EV ongoing-use footprint on par with natural gas (see above). With prudent EV charging done almost exclusively during off-peak hours, the effective ongoing-use EV footprint using coal-power grids can become nearly zero.

So much for “Coal Cars”.

How much off-peak capacity is there to spare, before nightly EV charging approaches the daily peaks and forces coal utilities to increase production? Thanks to the EV’s superior energy efficiency, quite a lot (see again why MPGe matters?). It is likely that coal will start phasing out nearly everywhere in the US, before we get to the point of lacking off-peak coal capacity to power EVs.To sum things up:

Reality Check 2 – summary:

Assuming today’s grid and typical EV charging patterns, on an ongoing-use basis, the GHG footprint of EVs is at least 20%-30% lower than that of the best comparable ICE hybrid. The difference can be far more dramatic depending upon region and EV driving/charging patterns.

An EV driver would have to engage in silly and almost deliberately wasteful driving/charging behavior, in order to incur a long-term ongoing-use GHG footprint worse than the best comparable ICE hybrid.

The only viable ICE competition to EVs on the ongoing-use footprint metric appears to be biodiesel vehicles. But unless you diligently fuel your biodiesel beast with waste-oil fuel, you will be assisting genetically-modified mass agriculture (for corn or soybeans), some of which comes with a footprint as bad as oil’s. This, besides the fact that diesel-exhaust fumes were formally declared by the WHO as carcinogeic. For these reasons, as well as its scarcity, I will leave biodiesel aside as a green ICE competitor from this point on.Note: the EPA just came out with an online ongoing-use GHG footprint calculator for EVs.The calculator compares the ongoing-use footprint of a specific EV you choose in a US zip code, to the new US ICE vehicle fleet average of 500 gCO2/mile. Differences between regions there are less dramatic than in other sources, but they too reach the conclusion that, e.g., the Leaf (even the 2011-2012, needless to say the more efficient 2013) has a lower ongoing-use footprint than the Prius (whose footprint is roughly half the national ICE average at 250 gCO2/mile) regardless of grid. Check it out: put in the zip of some coal-dominated region and see the numbers.

The bottom-line LCA figure from the 2013 EPA life-cycle analysis for EVs.
The bottom-line LCA figure from the 2013 EPA life-cycle analysis for EVs.

3. Life-Cycle Analysis

We are approaching the moment of truth: a complete LCA of the vehicle’s GHG emission footprint, cradle-to-junkyard. But paradoxically, this “truth” is anything but: already during the ongoing-use analysis, you’ve seen how variability, uncertainty and critical assumptions can tilt the entire numbers boat hither or thither. This is nothing compared to LCA.

And perhaps not surprisingly, since strategically-placed assumptions can tilt the entire analysis to your preferred direction, the assumptions about EV battery-pack footprint vary wildly. According to some schools of thoughts, it seems that EV battery packs are produced via a process of killing babies and burning pristine forests. I think here it will be simpler to begin from the Reality Check rather than end with it:

Reality Check 3a: recent estimates of the GHG footprint of EV battery pack production vary wildly, from 5 kgCO2/kg to 22 kgCO2/kg. The analyses showing EVs as somewhat inferior to the best hybrids on a life-cycle basis, used the high end of this range.

However, the most recent and most authoritative estimates are also the lowest ones.

The battery overhead has been the joker in the deck of EV-skeptic analyses and advocates. Climate Central’s report took the highest available estimate, and calculated the Leaf battery as costing 5.2 tons CO2 to make. That’s quite a hole to dig out of: on an ongoing-use basis, a Prius emits about 1 kg/CO2 per 4 miles driven. So according to the high-end battery estimates, a Prius can drive >20,000 miles just on the Leaf’s battery emissions! Or put another way, these 5.2 tons are about 70% of the emissions required (according to the same report) to build an entire 190-horsepower ICE car! Coupled with an assumption that Leaf batteries will need replacement every 50k miles on average, and giving the Leaf no discount for off-peak charging, Climate Central concluded that in some 2/3 of US states the Leaf would have a greater life-cycle GHG footprint than the Prius. Then they overstated their results (while downplaying their extreme assumptions) with the headline “Hybrids are greener than EVs…” Oh well. As a bonus, they made the same assumptions on the Tesla S, yielding 13 tons on the 60 KWh battery model — a setback from which it is nearly impossible to climb vs. the 40 MPG Lexus ES hybrid, especially when saddled with all the other EV-hostile assumptions in that report.Meanwhile… Dr. Jennifer Dunn and colleagues, at Argonne National Labs’ GREET team, decided to take a longer harder look at EV battery footprints. GREET stands for The Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation Model, and is considered one of the main global authorities on vehicular GHG footprint and other environmental impact. Dunn et al. list previous estimates ranging from 5 to 22 kgCO2/kg, examine in detail both the methodologies and the original problem itself – and come down hard on the side of the low-ball estimates (specifically 5.1 kgCO2/kg). The article itself is behind an academic paywall, but fortunately the supplement in which they scrutinize the battery-footprint estimates is free and open to all.

If you read that supplement, you’ll learn that the 22 kgCO2/kg number used by Hawkins et al. and Climate Central, has a strange history and a questionable methodology. It was taken from a previous paper, which in turn based its most critical numbers upon a 2005 study into batteries of photovoltaic systems (not EV batteries). The methodology used a “top-down” black-box approach, which in Dunn et al.’s words,

…may be double counting or their results may reflect boundary blur.

Dunn et al. strongly recommend the detailed process-based approach that details the energy consumption of each step. In particular, the lowest of these estimates comes from the EPA itself, and is virtually identical to the numbers finally adopted by Dunn et al.Using the Dunn et al. estimate, the Leaf battery’s footprint is suddenly only about 1.2 tons CO2. An ICE (the engine) takes more energy to build than the analogous EV powertrain (sans the battery), by about half a ton (these are Climate Central’s numbers). So using the Dunn et al. numbers for the EV battery instead of the inflated ones, the overall footprint of producing a Leaf-like EV become 8.2 tons CO2, vs. about 7 tons for a Prius-like hybrid and 7.5 tons for a plain-vanilla ICE compact. Not negligible, but not dramatic either. In fact, the difference is arguably smaller than the level of uncertainty about these estimates! Now, the Prius can barely drive 5,000 miles on its starting advantage, and ICE compacts can barely leave the parking lot on theirs.

Once off-peak charging is added to the mix, the Leaf’s life-cycle footprint is assured of being smaller than the Prius’ in all 50 states – turning the Climate Central headline on its head.  Even if we want to be conservative, and take a midpoint between the far-fetched, all-but-debunked 22 kgCO2/kg number and the more up-to-date ones, EVs still overtake the best comparable ICE hybrids rather comfortably, within 30-40k miles even under exclusively-fossil grids, once partial off-peak charging is accounted for. The Tesla S 60 KWh model, as well, now overtakes the Lexus hybrid fairly comfortably.

Reality Check 3b: under moderate assumptions, EVs have a lower life-cycle GHG footprints than the best comparable hybrid.

For example, a compact EV is manufactured with a 1-2 ton CO2 “deficit” compared to a compact ICE hybrid. Assume 2 tons, and assume an exclusively fossil grid in which the EV makes up only 1 kg CO2 per 20 miles driven, compared with the ICE hybrid (we are assuming moderately rational EV charging behavior). Then the compact EV will overtake the ICE hybrid, in terms of overall GHG emissions, within 40,000 miles.

Since we already know that present-day compact EV battery packs last more than 40k miles on the average, the EVs’ advantage is assured even given this relatively hostile scenario, even vs. the best comparable ICE hybrid.

How do we know current EV batteries last on the average at least 40k miles? Because with Leafs and other EVs on the road for nearly 3 years now, we have thousands of such batteries approaching or passing this milestone, and a very low replacement rate. Nissan felt good enough regarding its 1st-generation battery life data, that last summer it retroactivelygranted a 60,000 mile warranty on batteries of all its Leafs.But it gets even better for EVs:

  • The US grid keeps getting cleaner, shedding 8% of its footprint over just 2 years (2012 vs. 2010). As long as we can keep a global-warming denier out of the White House this will continue. Most states have policies or even mandates in place to keep improving the grid – policies which have withstood a concerted assault from ALEC. In 2012, for the first time ever, wind was the single largest source type in new power installations in the US.
  • The Dunn et al. report also examined the impact of recycling. At the moment the materials going into the Li-ion battery are not recycled. They show that recycling technologies proven at the laboratory level can reduce the EV battery footprint by half, bringing the overall EV manufacturing emissions on par with ICE cars.Given the generous government subsidies to the EV sector, governments surely have the leverage to spur the upscaling of these technologies.
  • Meanwhile, EV technology itself is far younger than ICE technology, and therefore is likely to improve energy efficiency far faster than ICE. Recently Toyota announced that their 2015 Prius will be 10% more energy efficient. They must be feeling the heat from something, because they have not bothered to produce even a 10% improvement cumulatively since 2004 (check it out here). Are they, perhaps, a bit worried about EVs upstaging them? Meanwhile, the 2013 Leaf is 16% more energy-efficient than the 2011-2012 version (you can check it out on the same link). EV makers have a far greater incentive to improve efficiency: it is the easiest way for them to improve the metric customers care the most about – the driving range.

Wait… Aren’t we forgetting something?

Tying together the many strands thrown open in the preceding paragraphs, one can reasonably safely assume that a typical new EV’s expected life-cycle GHG footprint is at least 10%-20% smaller than the best comparable new hybrid. But there’s one more element completely left out of the current LCA numbers game, an element on the ICE side. Its neglect is rather illuminating, because it indicates how slanted and warped the entire “Are EVs really Green?”game has been.

When thinking about oil and its problems, what is the first association in ordinary people’s minds? One that has dominated the global view of oil for 40 years? Right, war and conflict, especially in the Middle East where a plurality of the world’s oil is produced and roughly half of its remaining reserves sit, but not only there. Killing babies and burning pristine forests? That’s where you’ll find it – in the way oil gets from the ground to your pump.

Where are military energy expenditures, war and destruction in the oil GHG overhead calculations? Right now, nowhere. Dr. Burnham of ANL admitted that none of the GREET models account for it, and sent me to a team in UC-Davis he thought was working on such matters. The researcher there thought it was a great idea, and offered me to work on it with one of his students or post-docs, if he can interest one of them in the project. I still might do it… on my spare time of course. The rest of the world apparently has no interest in such numbers. Even the good guys: the GHG overhead tables from California’s Air Resource Board list Saudi-Kuwaiti oil as having one of the lowest GHG footprints in the world.

Well, almost no one is interested. There’s this 2010 Guardian piece, part of their “What’s the carbon footprint of…”, that quotes some 250-600 million tons CO2 as the Iraq war’s footprint. And finally, I even stumbled across a scientific article (by two agronomists from Iowa State U, Liska and Perrin) that tries to address exactly the question I was asking other researchers. They conclude that roughly 1/5 of US military emissions is related to “oil security”. Then they calculate that the overall GHG overhead for imported Middle Eastern oil in the US, due to US military expenditures and conflicts in the past decade or so, is close to 20%. When re-adjusted to reflect the share of this oil in the US total consumption, the number is reduced to a little over 2% for each ICE mile driven, regardless of petroleum source.

This study is a great start, but I think their calculation framework is wrong. Oil is not an American commodity, it is a global one. For example, when ongoing disruption in the Middle East due to the Iraq war caused an irreversible rise in world oil prices, it made tar sands and fracked shale oil more economically feasible. These sources carry a far higher GHG overhead, currently estimated at >40%. And the US, while being far and away the biggest military energy spender around Middle Eastern oil, is not the only one: the militarization of Iran, Iraq, Saudi Arabia, Libya and even a substantial part of Israel’s military expenditures can be attributed to oil. And we haven’t even counted the GHG associated with the tonnage of sheer destruction and the energy required to rebuild; the GHG cost of displacing and disrupting millions of human lives; etc.

Given that Middle Eastern oil accounts for ~30% of current production and ~50% of global reserves, I’d say that a 5%-10% overall global GHG overhead due to instability and disruption there and elsewhere (e.g., Nigeria or parts of the Amazon) is probably in the right ballpark.

And again, we haven’t yet counted the GHG impact of oil disasters and subsequent cleaning and rehabilitation overhead: Exxon Valdez, Deepwater Horizon, Lac Megantic, and oil spills great and small…  I’d say we can safely assume 7% overhead or more for Oil’s global problems combined.

4. Finally Putting it All Together

Ok, the final numbers. This is not a scientific article, but I’ve compiled and integrated a lot of information, and the numbers below represent the most likely footprint ranges IMHO, according to state-of-the-science knowledge.

Final reality check


  • The lower estimates of EV battery footprint are more reliable, but we’ll allow for something towards the middle (i.e., 5-12 kgCO2/kg-battery)
  • EV drivers charge roughly half their overall needs during off-peak hours
  • Compact EV battery packs last on average at least 60k miles, and Tesla S at least 100k miles
  • A 5% reduction in average electric-grid footprint over the life of the first battery-pack
  • A GHG overhead of 7% per ICE mile driven (not accounted for in conventional analysis) driven should be added, due to all Global Oil Troubles combined – military, conflicts, spills, disasters, etc.

We arrive at these ball-park figures:

  • A 2013-2014 compact electric-only EV takes 8-10 tons CO2 to make. Thereafter, it emits 150-200 gCO2/mile in fossil-based grids, and 50-100 gCO2/mile in renewable-dominated grids.
  • A 2013-2014 compact ICE hybrid takes 7-7.5 tons CO2 to make, and later consumes 270-350 gCO2/mile.
  • A 2013-2014 compact non-hybrid ICE car takes 7-8 tons CO2 to make, and consumes 400-500 gCO2/mile.

After 60k miles, we have:

A compact pure EV 11-21 tons CO2
A compact plug-in/extended-range hybrid EV driving 2/3 of its miles as an EV and the rest on gas (the current reported Chevy Volt average) 15-24 tons CO2
A compact ICE hybrid (non-plug Prius, etc.) 23-28 tons CO2
A compact non-hybrid ICE 31-38 tons CO2

Even if we assume an inevitable battery-pack replacement to the EV around 60k miles, its GHG debt will be smaller than the original due to the technology being more mature, and it will be recovered faster on average because the grid will be cleaner 60k miles from now. Conversely, while an ICE car is likely to survive well beyond 60k miles, its life beyond 60k miles is likely to be characterized by lower fuel efficiency and increased GHG overhead due to more frequent repairs.

On the Tesla segment, we have over the course of 100k miles (b/c Tesla’s battery-pack life is demonstrably longer)

60 KWh Tesla Model S 16-35 tons CO2
85 KWh Tesla Model S 18-40 tons CO2
The most efficient luxury ICE hybrid (Lexus ES 300h) 40-45 tons CO2
Non-hybrid luxury ICE cars (20-25 MPG) 60-75 tons CO2

So there you have it: from a GHG perspective, the first generation of mass-market EVs, despite the technology being far from optimized, is already greener than anything the ICE world has to offer.That said, even the best of compact EVs with the best current grid emits nearly 2 tons CO2/year on average, mostly due to production emissions. 2 tons is the global target for an entire family’s emissions if we want to stop global warming. Clearly, as said above, if one can get around without a car, and rent an energy-efficient one on the few occasions it is needed, then these tons are mostly saved. But for the vast majority of American families that still need a car to go about their daily errands, EVs as a major category appear to be the cleanest choice.

Moreover, given the multiple nearly-certain improvement paths such as:
– The grid will continue improving, and grid improvement affect both EVs’ ongoing-use footprint and its manufacturing footprint (the factories, after all, use electricity)
– Battery-pack life will continue improving, meaning we’ll be able to amortize the battery-production footprint over more years
– Battery material will eventually be recycled
– MPGe will continue improving

There’s reason to hope that within a decade we’ll be able to see EVs whose average annual life-cycle footprint is well under 1 ton CO2/year. ICE vehicles are not likely to come close to this.

I end with tentative answers to a few related questions.

Suppose you drive an old ICE wreck. Should you run it until it breaks, or get a new compact EV instead?
It will depend upon your specific circumstances, but just getting an EV alongside your wreck for everything except road trips, might “pay back” the GHG footprint of the new EV’s production thanks to the ongoing-use emission savings, within 10k-20k miles or so. This is pretty similar to the situation we faced a year ago; except that our wreck was about to simply die, had we continued to burden it with all our in-city driving. As a road-trip-only car it is surviving pretty nicely, touch rust 🙂

What about the EV + solar installation, popular among EV drivers? The connection might be more psychological than physical. It is a nice feeling to add an equivalent amount of produced clean energy, to the increased demand from your EV. However, rooftop PV systems are grid-connected as a rule. They supply the grid in daytime, whereas EVs charge at nighttime. In terms of sheer capacity, the US grid as a whole has plenty of room for many, many more EVs – 150 million of them. There might be local bottlenecks in EV-heavy regions, esp. if the desirable charging behavior is not encouraged, but I’m not sure PVs are what will be needed to mitigate them. In short, I suggest to separate the two decisions. If both an EV and a PV system makes environmental sense in your home and you can afford both – great. If only one is more sensible, choose that one (here in Seattle, I argue it’s an EV, b/c Seattle City Light already has excess green energy in the summer that it cannot always sell; in Phoenix AZ, Ground Zero of the “wilting Leaf” capacity-loss problem and blessed with eternal sunshine, solar makes way more sense than most EVs).

Last but not least…. why the differences between published LCAs? And why are my numbers slightly better for EVs than the best of them?

As said above, the Devil is in the details. The two most publicized EV-skeptic analyses used the inflated battery-production overhead now debunked by Dunn et al. They also assumed no grid improvements during the EV’s lifetime, and a peak-heavy EV charging pattern. To boot, they assume that the global Oil Troubles, precisely those which have given Oil its bad name, carry with them a zero GHG footprint. All 4 assumptions are demonstrably false.

The EPA report uses a lower and more realistic battery-production estimate, and (as far as I could tell) includes some allowance for grid improvements. The article text indicates that some sort of charging-scenario analysis was also applied, although a more complicated one than the simple halving of coal’s effective footprint as I did here (my halving was also based on a published report).

No LCA to date has penalized ICE cars for Oil Troubles. I decided to tack on a fairly modest 7% ICE overhead, based on the little published research into the issue that I managed to find.

As the numbers suggest, this additional overhead is not the decisive factor. But it is a nice and well-deserved ICEing on the cake of EVs’ overall footprint superiority (har har).


in the next diary I plan to deal with other environmental and social-environmental questions related to EVs.