The ANL study is quite old - the data in it is 10 years old, which is an eternity in this business. Here are some thoughts on battery cost:

A recent study Carnegie Mellon University argued that "plug-in" hybrid-electric vehicles, like the Chevy Volt, are too expensive. Are they right?

No. They assumed that the battery would cost $16,000 (or 1,000/KWH). As GM says, that's way too high. (Oddly, they also conclude that a plug-in with a 10 mile range would be better, because drivers would stop and charge every 10 miles!)

Similarly, $10,000 for the Volt's battery has been widely reported in the media, but we shouldn't rely on mass media! Really, no one knows how much the batteries cost. The $10K figure is purely speculation. Here's an example, in the CS Monitor. We see that it doesn't say $10K. Here's what the article says: "the race isn't over making a Chevy Volt battery designed to run 40 miles on a single charge that could (emphasis added) cost as much as $10,000." We can see that the reporter doesn't have a firm source for this cost figure.

Elsewhere, the article says: "Still others say that the cost of new battery power for PHEVs may drop faster and already be lower than what has been widely reported at perhaps $500 per kilowatt-hour or even less, says Suba Arunkumar, analyst for market researcher Frost & Sullivan.

"I do expect the price will come down to perhaps as low as $200 per kilowatt-hour when mass production begins in 2010 and 2011," she says."

Tesla's cost is $400/KWH - it's very likely that GM will pay $200-$300 in volume. The batteries won't be produced in large volumes for several years. They'll use less expensive materials than 1st Gen batteries; the larger format is much less expensive; and they'll have very, very large production volumes relative to most 1st-gen li-ion. Large production volumes reduce costs very quickly.

GM is pricing the Volt high purely to capture the early-adopter premium and the federal rebate - their official justification is that they're pricing in 100% replacement of the battery under warranty, which really isn't credible. We can expect the Volt to cost less than $30K with large volume production.

Is the battery too large?

Yes, they're only using 50% of the battery - a 50% depth of discharge (DOD) is very conservative. That means they have to use a 16 KWH battery to get an effective 8 KWH's. They could be more aggressive (and probably will be in the future), but they're very sensitive to the bad publicity that early battery failures would create.

Could they use a battery that allowed a deeper DOD?

No, there aren't any batteries on the market that are more durable as measured in charge cycles. Tesla's batteries aren't expected to last more than 400 cycles, and the Volt will do 5-10x as many. In theory, the Volt could have a smaller battery. That would mean a shorter range, which would still accomodate many drivers. That might more perfectly optimize costs, but then it wouldn't feel like a big step forward. It wouldn't feel like a real EV, with generator backup - instead, it would feel like an incremental hybrid. Both GM (for PR) and buyers want a large, step forward, I think.

Edit:

LG Chem uses lithium-manganese oxide (LiMn2O4) chemistry while A123 uses lithium-iron phosphate (LiFePO4) chemistry. LiMn2O4 has a limited cycle life and operates poorly as the temperature goes higher. The energy density is getting better as they incorporate more elements into the overall setup. But it isnt yet there comparable with LiFePO4 yet.

Why is this so? Better battery management system to avoid short-circuiting.

Could you expand on this? Are you suggesting that GM could have used a less sophisticated BMS if they had gone with LiFePO4?