http://www.transportation.anl.gov/pdfs/B/855.PDF This is a surprisingly approachable analysis (albeit already almost 5 years old, it seems fairly accurate (n.b. not in industrial battery production)). 20 pages on what it takes in energy to produce the precursors (primarily lithium carbonate), then what it takes to manufacture the actual cells, followed by an analysis of what we can recycle from cells which no longer hold a charge. I'm not saying fossil fuels are the way to go, but if you're driving around a Tesla in Nashville, odds are you're burning a boatload of coal in the process to get your charge from the regional power plant (blowing off the tops of mountains in the process to get that goal).

A software person would say that moving to EV's has the important benefit of separation of concerns.

If power generation is not in the vehicle, our national fraction of combustion generators can be reduced systematically. This is already under way; some US cities are already at 0.

That leaves storage on vehicle. The more EV's we have, the more incentive to find clean, high energy density batteries. This is not a static field either. Graphene is just one vaporware which might appear on the street shortly.

True, but this overlooks transmission inefficiencies. Inherent in moving power over distances is power loss (without superconductors). In that sense, local power generation is superior.

However, my understanding of the overall picture is that car engines (both diesel and gas) are so much less efficient than a power plant, that even with transmission, it's cleaner to switch to centralized power generation.

Gasoline also has large transportation inefficiencies. Unlike solar, every garden a refinery and oil well is impossible while every roof a solar pv is possible, even if 100% unlikely.

Are those inefficiencies actually so large? As they say, liquid hydrocarbons are a remarkably efficient method for keeping large amounts of disposable energy in a vehicle, currently difficult to match with e.g. batteries. And conversion of other energy forms to electricity involves some significant losses as well.

Line loss in US in 2007 was estimated to be 5% in 2013 (http://www.eia.gov/tools/faqs/faq.cfm?id=105&t=3)

For gasoline the number is supper tricky to determine. So here comes a back of napkin number

A fuel truck transports about 30,000 gallons (20,000-40,000). On average they achieve about 4-7.5 miles per gallon). So they use about 1% of their tankage for every 60 miles traveled.

This suggest to me that transportation loses for gasoline are in the same ball park number. Distribution and marketing is often claimed to be about 5% of the total, but I don't know where that number comes from.

All in all the difference is probably not that large, and I suspect the electricity one will drop faster than the gasoline one due to interconnect upgrades.

The paper you link seems very bullish on batteries. (“... we estimate that the energy use and greenhouse gas emissions associated with battery manufacturing make up only a few percent of a plug-in hybrid vehicle’s total life-cycle energy use. Further, the recycling of battery materials can potentially significantly reduce the material production energy.”)

Battery cost is still dropping at an exponential rate while energy density continues to rise. Likewise solar panel costs, and installation (especially at utility scale) is also coming down in price as the industry learns.

I’m no expert, but I predict that within 10 years it will be entirely practical (and much more environmentally friendly than diesel/gasoline, especially w/r/t urban smog) to have bus fleets, delivery trucks, etc. running on batteries.