I'm Not Crazy! Update on Electric Vehicle MPG

I will tell you that no matter how confidence in one has in his own intellectual ability, it's hard not to experience an "am I crazy?" moment when one reaches a conclusion different from everybody else's.  Case in point is my critique of the EPA's mpg numbers for electric vehicles.   The EPA's methodology strikes me as complete BS, but everyone, even folks like Popular Mechanics, keep treating the number like it is a serious representation of the fossil fuel use of vehicles like the Volt and Leaf.

Sot it was therefore nice to see a mechanical engineering professor independently make the same points I did in this Pajamas Media article. Also, my Princeton classmate Henry Payne, who often writes on automotive issues, linked my article at the Michigan View.

  • James

    You are not crazy and we can to a simular comclusion with our own back of the envelop calcuation. I'm sure it has its issues, but I think it is a reasonable assessment of the true MPG if such a number is even reasonable with electric vehciles.

    From DOE http://frwebgate.access.gpo.gov/cgi-bin/getdoc.cgi?dbname=2000_register&docid=00-14446-filed.pdf
    Tg = U.S. average fossil-fuel electricity generation efficiency = 0.328 (man, that’s much lower than I thought. I would’ve guessed 0.95)
    Tt = U.S. average electricity transmission efficiency = 0.924
    Tp = Petroleum refining and distribution efficiency = 0.830
    C = Watt-hours of energy per gallon of gasoline conversion factor = 33.705 kWh/gal
    Eg = (0.328 * 0.924 * 33.705)/0.830 =12.307 kWh/gal

    Now, this seems to be based only on thermodynamic efficiencies, without taking into account if the energy is coming from burning carbon, burning hydrogen, or splitting atoms.

    From http://www.epa.gov/cleanenergy/energy-resources/calculator.html:
    19.6 lb CO2/gal gas (just pure burning of gas, doesn’t include extraction/refining/distribution)
    1.58 lb CO2/ kWh electricity (national ave. w/o line loss)
    Ratio: 12.4 kWh/gal

    If we apply the efficiency values from DOE, 12.4*(0.924/0.83) = 13.8 kWh/gal
    Now, if you live in zip code 90210, it’s 0.763 lb/kWh (with line loss 5.33%), so the ratio is 30.5 kWh/gal

    As a sanity check, according to http://www.carbonfootprint.com/calculator.aspx, 10,000 kWh of electricity generates 6 metric tons of CO2, and 10,000 gal of “petrol” generates 87.65 metric tons. 14.6 kWh/gal

    The leaf is using 34 kWh per 100 mi. I suspect that includes charging losses, because the Tesla Roadster claims 21.6 kWH/100mi “battery to wheel” highway and 28 kWh/100mi “plug to wheel” highway (77.6% charging efficiency - that’s a steep hit right there). So, in terms of CO2 equivalent, the leaf is somewhere around 41 mpg for average Americans, which is just shockingly low, and around 90 mpg for Californians.

    Yeah, I spent more time on this than I meant to. Happy Thanksgiving.

  • Noah

    Nice to see these electric cars out just in time for winter. In a gasoline or diesel car, the heat to defrost windows and warm occupants is free. In an electric car, the same heat is a mileage drain.

  • D. E. Majewski

    I take it that you are referring to the coal powered cars versus the gas powered cars.

  • IgotBupkis, President, United Anarchist Society

    > even folks like Popular Mechanics, keep treating the number like it is a serious representation of the fossil fuel use of vehicles like the Volt and Leaf.

    Dude, the PopM and PopSci and Discover people all drank from the same brain-neutralizing Leftist Kool-Aid Acid Test decades ago.

    Just look at their breathless reporting on AGW that lacks any sort of critical or skeptical attitude.

    OF COURSE they're going to parrot the Libtard line on anything GuhREEN.

    I stopped paying any attention to Discover when they ran this breathless piece about 2 decades ago wherein they took "art" produced by "animal artists" to "art experts" for "their opinions"... without telling them the art came from animals.

    The variously positive responses were taken (seriously!) as examples of how animals could do art, just like humans, rather than either a commentary on the value of "art experts" for the most part, or, more likely, a serious commentary about the state of so-called "Modern Art".

    I mean that the latter notions, which at least are worthy of mention even IF you support Discover's base premise, never came up in the article at all. Either it never entered their tiny widdle minds or they were too busy selling the idiocy to dare think that anyone reading their twaddle might be capable of independently arriving at either of those notions... Exit me from readership, stage left.

    PopM and PopSci both subscribe to the same school of pseudo science as Discover. While one might find some actually useful, reliable information therein, it's accidental, and totally incidental to the real underlying purpose of the magazines -- to make people -- particularly leftists -- able to believe they are informed about science.

    Ask a typical reader "What is Science?" -- I will almost guarantee you will NEVER get anything close to "A set of methods for determining, with some reliability but no absoluteness, the truth value of an assertion".

    Science isn't a method to the Lefty -- it's a set of factoids and absolutely truthful assertions Which Cannot Be Doubted.

  • Tim

    He made the same mistake you did. That doesn't somehow make you both right.

    Here's a different way of showing why you are wrong:

    If you accept the premise that the EPA mileage numbers need to be adjusted down using some factor to account for the losses of generating electricity, you've now created an unequal method for the EPA window sticker numbers. Why? Because there is no adjustment or correction factor applied for losses in going from the well head to the pump for gasoline. So, if you really want to apply this 'correction' for electric and plug-in vehicles; you need to figure out the correction for combustion vehicles, too.

    That same thing applies to the tailpipe emissions. If you think the EV needs to have, on the sticker, some accounting for its system-level CO2 emissions; then you need to add the CO2 accounting for exploration, pumping, transport, refining, finished gas transport, etc. of gas.

    If you don't, the numbers on the window sticker aren't created with the same procedure; and therefore are an unequal basis of comparison.

    By the way, as I may have mentioned before, the EPA numbers are *measured* numbers. The city numbers are measured on a defined drive cycle with the vehicle on a dynamometer. The highway numbers are measured on a defined "road" course, but actually performed on a closed track. They use calibrated fuels and actually measure how much fuel is used for the defined cycle -- giving the rated miles per gallon.

    The purpose of the EPA numbers are to provide information to the consumer. It tells them, roughly(1), how far they can expect to go on a tank of gas. It provides a means of comparing vehicles to each other in terms of fuel efficiency.(2)(3) It is not designed, nor is it intended, to account for the *total* system cost.

    The DoE/NHTSA number, (a/k/a the CAFE program) is designed to account for the total system cost. It uses the EPA numbers as a baseline for the vehicle's fuel consumption, combines city and highway driving numbers on a proportional basis to get one number; then applies the efficiency correction factor to account for raw material to vehicle losses.

    In short, two different questions need two different methods to get answers.

    Now, if you think the question that the EPA numbers answer should reflect the system cost; then you need to change the question and have *all* stickers reflect the new answer. But then the EPA sticker number will be further away from reality as cars are going to go further, and be much more efficient than their rated numbers, because the system costs are external to the consumer experience of paying for fuel.

    And, in the end, that's really the point of the test methodology that generates EPA's numbers. It's supposed to reflect the *consumer* experience. If I buy an IC vehicle that gets 50 mpg and has a 10 gallon tank; I would reasonably expect to be able to go 500 or so miles before I refuel.

    How can I comparison shop that car with a Volt or a Leaf, if the stickers on the EVs have externalities included? Take the Volt. On a full battery charge, in mixed driving, it gets 60 mpg. So, given traffic mix, driving style, etc; I could reasonably expect to get that ~370 mile range. If you adjust the Volt numbers down to account for all those externalities; my range would be much better than stated on the sticker.

    The real issue isn't that the EPA test doesn't reflect externalities; it's that there is no place to reasonably signal all those externalities to the consumer. The CAFE number is already too abstracted(4), putting the externalities in the CAFE number would make them too abstract, and the window stickers are already approaching information overload. (5) You could signal with fuel taxes; but that would require a rational energy policy.

    (1) I say roughly because the EPA drive cycles don't really account for how people actually drive the car. It also assumes a fuel quality and ethanol content (0%) that you may not actually burn.

    (2) It actually doesn't do a good job of that, either; because MPG is non-linear. See http://www.mpgillusion.com/ for all the details.

    (3) One other bit of shenanigans. The Volt got classified as a compact; but the Leaf is somehow a mid-size car. So if you are using the 'compared to' part of the sticker, you have to be careful that you are comparing within a class.

    (4) It makes assumptions about city/highway driving mix which distort the number from what a particular driver's experience might be.

    (5) There's the car sticker; with the standard and optional feature list, the price breakdown, point of origin, VIN, and EPA label; there's the bumper rating sticker; there's the US content sticker. At a certain point, it just becomes noise.

  • markm

    Shorter Tim: Because one number neglects an adjustment of around 15%, you should neglect an adjustment of around 70% in calculating another number.

  • Tim

    @markm: I think you missed my point.

    Here's the short version:

    One set of numbers are used for a different purpose than the other set of numbers.
    Because of how one pair of numbers are generated and used; you properly *can't* include any adjustment factor. The other number (note, it isn't a set) is generated differently and used differently; and therefore *does* include the adjustment factor.