>One less thing for anti-hybrids to whine about...

The test route is the same, but the calculation methods differ slightly. I don't understand your objection. For the case of a non-hybrid vehicle, you just omit the compensation factor for SoC change.

The test would have to be infeasibly long to reliably make SoC insignificant (ie not affecting a 3 significant figure result). What if it's a plug-in hybrid with a 50 mile range?

And how do you subject an electric vehicle to the same protocol as a petrol car? You can't measure the CO2 output and calculate a mpg figure

The test procedures are something like:

1) For a fuel-only car, run the test, measure CO2 (and other) output, calculate l/100km.
2) For a non-plug-in hybrid, do the test as 1, but if the SoC differs between the start and end of the test, then adjust the result
3) For an electric car, fully charge the car, run the test, then recharge back to full, measuring the power drawn from the charger to give a Wh/km figure.
4) For a plug-in hybrid, perform the test twice, as per 2 and 3 to give separate figures for fuel and electricity consumption.

There are extra conditions about how to treat switchable hybrids (eg the EV button would have to be left unused).

 

This entire business of revising EPA testing is so mired in self-interest and general ignorance to have lost any utility.

If I had my way, I would have one test that simulates a lead-footer driving in LA on a hot and humid day, and a second test of gentle driving in a temperate climate. Smart people would understand their FE will likely be within that range, depending on climate, roads, and driver habits.

 

The EPA MPG estimates I've seen have all been 2 digits.
Also, their tests cannot infer mileage (i.e. from CO2) but must measure it.
(Measured miles per measured gasoline)
The EPA certainly could not get away with measuring the battery charge before and after and adjusting the MPG accordingly.
I can tell you that, from my mechanical engineering training, I know that to be invalid.
Combining tests 2 and 3 would also not be legitimate because that would not reflect real-world driving conditions.
The only legitimate way to eliminate the SoC difference (if it is determined that it could be significant) would be to start and end the test at the same SoC level (repeatedly).

My understanding is that you could go about 3 miles on EV from a full battery.
Presumably, the SoC difference between the start and end of a test would power the car for significantly less than that and could be probably be ignored without affecting the result unless the test is very short (say less than 10 miles), which I doubt.

As you observed, MPG is not applicable to electric vehicles.
Miles per kilowatt-hour would be correct for these.

Plug-in hybrids are essential dual fuel vehicles and, as such, MPG is also not applicable to them because energy, in the form of electricity is added.
Computing a single apples-to-apples number for these would be problematic.
I think the simplest for the average American to understand would be cost per mile (siting basis electric rate and gasoline price).

As we eventually move away from gasoline we'll probably be using cost per mile across the board.

 

Actually, you're right. The basic UNECE test give l/100km to 2 significant figures. But in the UK they then convert that to mpg with 3 significant figures (totally spuriously for high mpg cars like the Prius).

In the UNECE case they don't measure the charge directly, they monitor and total the currents to and from the battery during the test. If you are interested in the exact calculations used, see http://www.unece.org/trans/main/wp29/wp29regs/r101r2e.pdf.

They don't combine 2 and 3, they quote the figures separately - ie it has a fuel mpg figure AND a separate electric efficiency figure. There is also an official "electric-only range" test - something there hasn't been any proper standardisation on so far.

Given those 3 numbers, then a potential purchaser can estimate the cost efficiency to them of a plug-in hybrid - something that will vary massively depending on the average journey distance.

The test distances currently used are actually pretty short. In the UNECE test, the urban cycle is 4km or so, and the extra-urban cycle is 7km. I wouldn't have thought the EPA test is terribly different. So being able to go just an extra mile by depleting your SoC would be massively significant.

It certainly is applicable - you want to know how much fuel it will consume if you don't plug it in or once the plug-in charge is exhausted.

I agree that's easiest to understand, but to calculate it, I think you need to generate the basic info first - eg given 5.0l/100km fuel efficiency, 120Wh/km electric efficiency and 30km electric range, you could then calculate different cost/km figures for say, 5km commutes, 20km commutes and 50km commutes.