The external electricity, as I pointed out doesn't make a huge difference (0.31 kWh per gallon or ~5% of the total energy 6 to 7.5kWh losses from refining).
I missed that you already had a number for that, and didn't expect it to be that low. (I didn't follow those previous discussions at that level of detail since, I think, at that time I already thought it would be better to use solar energy in combination with EVs as the more valid scenario. Recent price developments in solar seem to confirm that view. See also my post from Nov 18, 2009 in this thread).
But, if you go that far, why do you stop at that point (at refineries) instead of going further and including electricity that is available in the night (off peak), for example?
The reason I use 40% average for natural gas is to keep in the spirit of the analysis (which looks at refining efficiency nationwide, not just the newest, most efficient refineries in operation).
I don't see why you would think that using averages including old plants would be more in some "spirit". If 60% efficiency to process that NG is possible, then I don't see a reason not to actually do this. The idea isn't to keep wasting the energy except to use it for EVs. I'm talking about the possible future reality, not about a restricted thought experiment.
Average efficiency of fossil fuel plants in the US (2009):
Coal - 32.8%
Petroleum - 31.2%
Natural Gas - 41.8%
http://www.eia.doe.gov/cneaf/electricity/epa/epat5p3.html
(efficiency in percentage can be figured by dividing 3,412 by the associated heat rate in the link above; see link below for details)
http://www.eia.gov/tools/faqs/faq.cfm?id=107&t=3
Notice natural gas has been steadily improving, while coal and petroleum plants have stayed the same; seems to reflect the trend that most new plants built are natural gas plants.
As far as I can tell from the ANL study right now, coal isn't used, but very much something called "still gas"...
The ANL link I gave [...]
That's the same or similar one I referred to in Nov 2009 in this thread.
In my Feb 27, 2011 post in this thread I added a link to the updated 2010 version of that report.
[...] should have most of the necessary data to figure out the total electricity that can be made from all the feedstocks to make gasoline. The only missing data is the conversion from the feedstock to electricity. I've tried to do a rough analysis (back when I first posted that ANL link) taking into account the external electricity, and also the electricity that can be made from the natural gas, steam, and coal feedstocks, but ended up with only about 1kWh.
I'm not sure what you refer to as 1 kWh at the end, but I can now see how you would arrive at 2.4 to 3 kWh, even though I would still arrive at a larger number (probably closer to 4% than to 3%, at least if "still gas" could be processed as effectively as NG).
BTW, if the average of US NG plants is 40%, I'm wondering whether the older ones should be upgraded to 60% as soon as possible.
And once more, it should be pointed out that the long tail pipe argument is most often mentioned in the context of CO2 production, and in that context, the number of around 6.7 kWh for the amount in refineries which corresponds to CO2 production of electricity for EVs, still seems the closest estimate.