I think oil sands do get an extra 20% hit, and Glenn includes that in his analysis. But I also gave marginal electricity an efficiency pass by using 40% which is closer to grid average.
The 40% (modern technologies are higher) do not play into the CO2 numbers either, since that's already accounted for by the 610 g-CO2/kWh which he uses for NG.
In California, night time marginal emissions are (according to the 2009 study) around 575 g-CO2/kWh if you need 4 hours of charging (or 568 g-CO2/kWh if you need 2 hours of charging, and choose the best time in each case). (While NG tech is improving). His NG value for the Leaf assumes higher fast-charging losses of 25%, if instead you assume night charging with NG that is according to his numbers 10% instead.
His best-case calculation is for fast-charging NG: 0.34 kWh/mile * 1.25 * 1.06 * 610 g-CO2/kWh = ~274.8 g-CO2/mile.
However, in California, which has in the order of
half the EV sales of the US (so far), night time charging (most common) will be, according to *his* formulas, 0.34 kWh/mile * 1.1 * 1.06 * 575 g-CO2/kWh = ~228 g-CO2/mile.
That's
228 g-CO2/mile for the Leaf in CA, vs.
280 g-CO2/mile for the Prius, according to his numbers for tar sand. Now if it is correct that previous studies "have vastly underestimated the carbon footprint of the Canadian oil sands", then that difference is even "vastly" larger than that, and even worse for the Prius.
Probably the numbers will be similar for other states that use NG for marginal power, such as New York State. I haven't found any data yet about the marginal power of other states, but his assumption that coal is being used for night charging, flat-out, turns out to be wrong.