Author profile picture

A rebuttal to the German research article:  Fossil fuel Engines, Wind Engines and Diesel Engines. What does the CO2 balance sheet reveal?”, written by Christoph Buchal, Hans-Dieter Karl and Hans-Werner Sinn.

A recent opinion piece by Buchal, Karl and Sinn, claims that a Tesla Model 3 emits 11% to 28% more CO2 over its lifespan than a Mercedes C220 diesel car does. A correction of the errors in their calculations reveals that the electric vehicle (EV) actually emits 63% less CO2 than stated.

By replacing this inconsistent NEDC test with real-world values, diesel emissions revealed an increase of 80 grams of CO2/km driven, while electric emissions showed a CO2 increase of just 15 g.

By replacing an outdated study on battery production with new research and by also taking into account that the Tesla Gigafactory runs on renewable energy, battery CO2 emissions are thereby reduced from 85 g to 33 g. By increasing battery lifespan up to 300 thousand kms, battery emissions are further reduced down to 16 g.

Substituting current German energy blend data with the vehicle’s energy blend over a lifespan, reveals a reduction in CO2 emissions from 97 g down to 66 g for an electrically driven vehicle.

The end result is that CO2 emissions from the Mercedes C220 diesel show an increase from 141 g up to 221 g, whereas CO2 emissions from the Tesla Model 3 show a decrease from 167 g down to 83 g.

Recently, Buchal, Karl and Sinn (aka BKS) published a much-cited study claiming diesel-powered cars emitted less CO2 over their lifetime than electric vehicles. My impression was that it was basically a list of over-the-top worst-case assumptions about the EV, in an effort to make diesel look good, produced by people with limited knowledge of electric vehicle research. I pointed that out on twitter and many others have also contested their publication.

Yet now the authors are back, with a blog post titled Erläuterungen zur Studie “Kohlemotoren, Windmotoren und Dieselmotoren. Was zeigt die CO2-Bilanz?” (‘Explanatory notes on the study: Fossil fuel Engines, Wind Engines and Diesel Engines. What does the CO2 balance sheet reveal?’). Wherein they essentially double down on everything that they have written.

It would be tempting to disregard them, because although the writers are respected scientists in their respective fields, they have never researched electric vehicles. Therefore, as far as I’m concerned, these are the non-peer reviewed opinions of laypeople. However, Hans Werner Sinn is a famous economist in Germany, and their conclusions were remarkable, subsequently they gained a lot of traction in the popular press. As I am a scientist who researches this at the Eindhoven University of Technology, I feel responsible for keeping the public debate free of obvious errors. Subsequently, I have written this blog post in order to correct any major inaccuracies.

More from Auke Hoekstra on Innovation Origins? Read here.

First, let’s quickly set the scene. A long-range Tesla Model 3 EV and a diesel-powered Mercedes 220 are going head to head: which car emits more CO2 over its lifetime? Problems for the diesel are: it emits more in real life than in the lab plus you must not forget to factor in the production of diesel. Problems for the electric car: manufacturing the battery emits CO2 and the German electricity blend is dirty. Which car will win?

BKS made a convenient list of criticisms in their blog post which they have doubled down on. I will follow up on their questions. I translated the questions into English and abbreviated their answers. (You can look at their blog post for their answers in full.) Then I added my own answers and explanations. I’m hoping you’ll want to follow this all.

1. Did we overlook the importance of EVs with regard to air quality in the city?

BKR’s answer: no, and we do not mind electric city vehicles at all, however we did make a comparison with regular cars.

My answer: you overlooked ‘Dieselgate’ and 5,000 deaths per year.

Consequently, someone complained to BKR that diesel cars are emitting deadly NOx in the city, and they chose to ignore that. (Diesel cars are estimated to cost 5,000 deaths per year) BKS basically replies that they are fans of small electric city cars and electric scooters, but that they are comparing long distance cars here. Well, I’m a fan of small electric city cars and of electric scooters too! (And I like bicycles even more because I’m Dutch.) But that does not alter the fact we are comparing cars here so let’s stop deflecting from the subject at hand.

Also, most people are aware by now that diesel engines have a problem. They look great in the lab after carmakers are done fiddling the test, but on average, diesel cars pollute 4.5 times more than they are allowed to in real life. (Mercedes is average.) BKR could have mentioned some of that.

2. Did we overlook the usefulness of car batteries which charge on wind and solar grids??
BKR’s answer: we did mention green methane and hydrogen but you can’t use car batteries for seasonal storage.

My answer: you seem unaware that EVs are extremely useful when it comes to daily grid management.

auke hoekstra diesel vs. electric

The picture shows some attributes of one of our models (page here, the film is in Dutch, unfortunately).

Explanation: it’s my job as a researcher to model future electricity grids. Among us (energy modelers), it’s well known that you basically need two new types of storage to deal with solar and wind: day-to-day and yearly. Saying EVs are useless as seasonal storage is beside the point and nobody in his right mind uses batteries (inside or outside a car) for that. But EVs are a godsend when it comes to daily storage and variable loads. That means they can use wind and solar when all other options are gone. This improves business prospects for wind and solar, and strongly reduces the additional amount of stationary storage batteries that you have to manufacture and buy.

What variants of the Tesla did we compare?
BKR’s answer: we took the one within the range closest to the diesel vehicle.

My answer: this is the biggest battery in this price range. Why not take an average one?

Explanation: I’m not saying that this is an error. But it’s certainly not average. Maybe it would have been better if BKR reminded people that the 75 kWh battery they chose, represents the biggest battery in this section and that the Model 3 also comes with a 50 kWh battery (335 km real range).

3. Why didn’t we use WLTP?
BKR’s answer: because Tesla did not yet have a WLTP (Worldwide Harmonised Light Vehicle Test Procedure) value.

My answer: if we take the real-world energy usage into account, the CO2 emissions from the Mercedes increase by 80 g and become 221 g. The Tesla only increases from 15 g to 182 g. The diesel is already lagging behind by 40 g after the first correction.

Explanation: in order to make it possible to compare cars in an objective fashion, the world uses standardized tests. In Europe it used to be the NEDC (New European Driving Cycle). But the difference between the NEDC and concrete reality continued to grow, mostly because car makers are in charge of the tests and they learnt to apply more and more tricks. On average, modern vehicles (since 2015) emit 40% more CO2 in reality than in the NEDC test. How do we know this for sure?

  1. There are dozens of studies by serious research institutes (like the ICCT report above) which compare NEDC with official road tests and they all show the same: the gap between NEDC and actual usage is becoming wider and wider.
  2. In America, it’s not the carmakers but an independent organization (the EPA) that tests CO2 exhaust emissions; and in the USA, the numbers are (surprise!) quite realistic.
  3. There are many sites which have users keeping track of their energy use. E.g. the German

The successor to the NEDC is the WLTP (Worldwide Harmonized Light Vehicle Test Procedure), and it is somewhat closer to reality. But there is already evidence for cheating, which is to be expected, since tests are still done by manufacturers themselves. Basically, everybody in the industry knows you have two types of tests: tests done by the manufacturers themselves (NEDC and WLTP) and tests done by independent bodies (e.g. EPA and ICCT).

Therefore, did BKR really want to compare realistic values? Or did they want to compare incorrect numbers that make diesel look good? Maybe they were unaware of the problem at first, yet their unwillingness to use more realistic values, after this has been pointed out to them, remains troubling.

Using the more realistic energy usage values from, the Mercedes C220 diesel CO2 emissions jump by 80 grams from 141 grams of CO2/km to 221 grams of CO2/ km. The Tesla Model 3 also increases, but by a mere 15 grams from 167 to 182 grams. We could also take new road tests done in Europe, or the figures from the American Environmental Protection Agency, and the results would be similar. Approximating real-world usage is only hard if you are a car manufacturer who wants to pretend that your car uses less than it actually does. Taking real-life energy usage into account, immediately puts diesel energy at a disadvantage. But we are just getting started!

5. Did we factor in too much CO2 emissions for the battery manufacture?
BKR’s answer: of course not! We used the really good meta-study from Romare and Dahllöf (the Swedish study).

My answer: The Tesla Gigafactory does not emit 170 kg of CO2 emissions per kWh of battery power, but rather 65 kg. This lowers the battery CO2 emissions per km from 85 g to 33 g. The Tesla now emits 130 g versus 221g for the Mercedes.

The Swedish study is the go-to source for EV sceptics because of their unrealistically high values for CO2 emissions for battery production. Based on this, BKR assumes that between 145 and 195 kg CO2 (let’s average it at 170) will be emitted for every kWh of battery power produced. They based that on a study that managed to rile up a lot of people a few years ago and which has been amended several times since. Here’s one example in Handelsblatt. The authors also felt compelled to point out that battery manufacture is rapidly advancing, and that their study should not be used to predict future battery emissions like BKR are currently doing. This is, of course, to be expected: as batteries get lighter, they use less material and as competition grows fiercer, bringing down energy expenses is vital.

Elon Musk called the Swedish study ‘clueless’ and pointed out much less energy is required per kWh of battery power, and his factory runs on renewable energy anyway. Now you might dislike Elon Musk, but he does have the biggest battery factory on earth and more specifically, the one where the Model 3 is made which BKR is talking about. You know, the factory that will be covered in solar cells, and with enough windmills next to them to produce all the energy they need. Tesla has told the world repeatedly that Tesla already buys 100% green energy and that the factory will only use renewable electricity, specifically built for the factory in 2019. (Of course, in “Elon Time” that might not be until 2020 or 2021.)

auke hoekstra diesel vs electric

Partly as a reaction to the Swedish study, FFE did a follow-up study titled “Carbon footprint of electric vehicles – a plea for more objectivity“, incorporating the authors’ reactions after the study and looking towards the future. The study concludes that 6 kg per kWh is a better estimate right now. However, this can get up to 65 kg of CO2 for a factory running on renewable energy, such as the Tesla Gigafactory. I might add that mineral extraction could also be done with electric vehicles and other electric machinery running on renewable energy and this would again slash that 65 kg per kWh number in half. Plus, if we take the second life of batteries and recycling into account, things look even better. A study from February 2019 by Maeva Philippot et al., (from the excellent research group run by Joeri van Mierlo at the Vrije Universiteit Brussel) concludes that “Optimizing the process by reducing the electricity consumption during the manufacturing is also suggested, and combined with higher pack energy density, the impact on climate change of the pack manufacturing is as low as 39.5 kg CO2 eq/kWh.

Now, why in heaven’s name would BKR use this oft-corrected study, and not a more recent one? I won’t speculate but let’s just say the numbers are outdated. For the Tesla Gigafactory, the best estimate is not 145 kg – 195 kg but 65 kg of CO2 per kWh of battery power. (Which also correspond to industry sources of mine.) And in the future, that number might easily go below 40 kg.

6. Should you take into account battery production which uses green energy?
BKR’s answer: no, you have to take the average, otherwise you just end up subtracting green electricity from somewhere else.

My answer: if you are making a comparison with a Tesla Model 3, you should certainly do so.

We have already covered this above. Tesla claims to already manufacture batteries using green energy from wind and solar sources. (By the way, the Nevada combination phased out almost all coal between 2004 and 2015.) They are currently on track to get all energy within a few years from their own combination of solar, wind and batteries. Not counting that as being manufactured using green sources of electricity, would be ludicrous.

And let’s take the reasoning of BKR to its logical conclusion: you can buy green energy; you can buy stakes in a windmill next to your place, and you can even buy solar panels for your roof. It’s all to no avail. Apparently, only nationwide policies can make a difference and there is nothing anyone can do as an individual. I find that a very gloomy, unempowering, status quo-friendly kind of reasoning. I’ll count my own solar panels as self-induced green energy, thank you very much. But apart from that discussion, the Tesla Model 3 will soon be produced using their factory’s own renewable electricity. So, not counting that is just wrong

7. Did we exaggerate the CO2 emissions of the German electricity combi?

BKR’s answer: no we didn’t.

My answer: the best conservative estimate on the lifespan of this vehicle is 375 grams per kWh, not 550. This lowers Tesla emissions by 31 g. The Tesla currently emits 99 g and the Mercedes, 221 g.

In the quest to reduce CO2 emissions, many things are happening simultaneously. Two very much interconnected things are the transition from diesel cars to electric vehicles, and the transition from fossil fuels to gas, solar and wind. The one trick that literally every EV opponent uses, is to assume that nothing will happen on the fossil fuels front. So, not only do they use the current high CO2 German combination, but they also assume the combination will not become any better during the vehicle’s lifespan.

That’s a strange supposition as there are German laws about that, requiring 40% renewable electricity in 2025, 55% in 2035 and 80% in 2050. The coalition accord of the current CDU-CSU-SPD government even has a target of 65% by 2030. A German energy modeling colleague recommended this study for the German ‘Stromsektor 2030’. It estimates the grams per kWh in 2030 in three scenarios:

  1. BAU (business as usual): 413 gr
  2. KA (Kohleausstieg/Fossil fuel phase-out): 335 gr
  3. KA65 (Kohleausstieg/ Fossil fuel phase-out) with 65% renewable electricity, the current government policy, 310 gr

Now we can argue about which number is most realistic. Nevertheless, assuming that the current figure of 550 g will stay the same over the lifespan of the vehicle (which BKR does), is clearly simplistic and wrong. I propose we imagine that the car will be driven for 17 years (it’s 19 in the Netherlands) up until 2036. I estimate the mix in 2036 will be below 200 g of CO2 per kWh. If we average out the difference between 550 and 200, we get 375 grams. I’m not saying it’s a perfect figure; however, I do think it is a lot more realistic than the amounts assumed by BKR.

Since the Tesla uses 0,177 kWh per km (see step 4), reducing the energy combination from 550 g to 375 g, lowers the emissions per km for electricity from 97 kg (= 550 x 0.177) to 66 kg (= 375 x 0.177).

8. How long does an electric battery last?

BKR’s answer: car manufacturers claim unrealistic numbers like 300 km, but we used a more realistic 150 km.

My answer: the battery will last over 600 km, but let’s ignore second life and recycling and assume it is scrapped along with the car after 300 km. This halves battery emissions to 16 g. Now the Tesla emits 83 g and the Mercedes, 221 g.

First, let’s look at what 150 km actually means. If a Tesla with a 485km actual life range were to drive 150 000 km (as BKR proposes) that would constitute 310 cycles.

In reality, modern batteries like the Samsung battery in the BMW i3 are rated at 4600 cycles. For the Tesla Model 3, that would constitute 485 x 4600 = 2.2 million kilometers.

Performance is less than this in real life. However, the quality of batteries is rapidly evolving. Real world tests of Tesla batteries by actual owners show that they will last anywhere between 500 km and 1000 km. If BKR are really interested in battery degradation, I propose they start by reading some 2018 sources here, here, here, here, here, here, here and here. Bottom line: battery developments are happening incredibly fast and tomorrow’s batteries will last even longer than today’s batteries, which already last about ten times longer than what BKS are currently assuming.

Pinning down a more accurate figure is much harder. An electric car’s motor and battery will probably outlast the rest of the car. Consequently, after how many kms will a car be put to rest? Hard to say, yet we know Tesla Model 3 cars are driven more kms than average because of their low cost per km, thus the comparison with diesel cars is appropriate. In the Netherlands, 300 km is a low estimate for a diesel, and I would guess people in Germany do not drive any less. Therefore, I propose 300 km as a conservative esitimate.

9. Don’t we have to include the diesel motor production?
BKR’s answer: Nah, it’s all relative.

My answer: Nah, it’s all relative. We agree!

10. Do electric cars come off badly in our study?
BKR’s answer: on the contrary because diesels have more range, whereas chargers use energy, heating EVs costs energy and we even took diesel production into account.

My answer: by correcting the errors in your study, the energy use of the diesel car increased from 141 g up to 221 g, while the emissions of the electric vehicle decreased from 167 g down to 83 g. Subsequently, the EV went from being worse by 10 to 25% in your study, to actually performing better by 62% after correction.

Strange that BKR want to mention once again they did not forget to factor in the production of diesel. Let me just say: I don’t want to make an issue; of course, you have to include diesel production and unsurprisingly you took a pretty low estimate. Energy for cooling of chargers is negligible: it’s loud but doesn’t use that much energy compared to the 100kW that gets delivered. Heating of EVs takes some energy but it’s pretty negligible too. BKR were right to ignore those as well.

So, let me sum up the answer to this question by looking at all BKR’s questions (every line corresponds to the question with the same number):

  1. BKR ignored air pollution.
  2. BKR ignored the usefulness of EVs in stabilizing wind and solar grids.
  3. BLR used the largest battery in this price range.
  4. BKR took unrealistically low energy usage into account, which gave diesel an advantage.
  5. BKR measured unrealistically high battery production emissions, which was a disadvantage to EVs.
  6. BKR disregarded the fact that Tesla batteries are made using green electricity.
  7. BKR disregarded the fact that EV electricity consumption will become greener over a lifetime.
  8. BKR underestimated the lifespan of a battery.

So yes, I think it is safe to say electric cars came off rather badly in this study.


UPDATE:  The first version of this article erroneously stated MtCO2/a instead of gCO2/kWh for the KA and KA65 scenarios and referred to European laws on renewables, where it should have been German laws on renewable electricity. This has been corrected and the text directly below it has been updated to reflect this correction.