More Thoughts on Energy Storage and the Powerwall: Power and Energy

My meandering thoughts were amplified a bit and I was taken to task but I think came out OK. And it brings up important points that deserve further text. TL;DR: The power rating of the battery matters, and a 3 hour discharge basis requires a cost of less than $220/kWhr given relatively expensive gas to compete in solar firming.

A battery, per cycle, needs to be significantly cheaper than the spread of day/night energy pricing if we’re talking solar vs. grid, and generally better than peak vs. off peak pricing. Both Tesla Powerwall batteries, based on the difference in NYC peak demand pricing, will find many customers if generators are verboten. But generators are the real competition here. Generacs at a home scale, efficienct fast ramp GT’s at the grid scale. I’m a battery guy, I work on batteries for a living, and we like to think that gas turbines are mature and tapped out. The truth is that for a given $/kW rating they’re getting better all the time. More efficient overall and over larger power ranges, and faster to spin up and down. So we need to compare batteries on a power to power basis with gas fired anything.

The Tesla batteries are 5 hour and 3 hour batteries, respectively. In all batteries energy and power are coupled, so as the table above shows the longer the discharge of the battery, the more the power costs. A home generator costs $4,000 for 20 kW (ref), or $200/kW, so to match sustained output that you’d need 10x the batteries listed above.

This is why early adoption of batteries on the grid has been for high power (e.g. frequency regulation) applications using lithium ion batteries built for power. The math works nicely here: if a 10 kWhr battery can be discharged in 1/2 hour, at a cost of $350/kWhr this is equivalent to a generator that costs $175/kW, or better than our generator in point 2. Here is an example of a 32 MW/ 8MWhr battery, or a battery capable of giving up all of its energy in 15 minutes. It doesn’t have to, it can run slower. But it can. Of course, if we spend all that energy in 15 minutes and need more we are SOL.

But let’s say we know how much energy we need for how long a priori. With the above relationship we can determine how much a battery can cost, AC to AC, against a generator. Let’s assume we’re going against an industrial generator, not a home use unit, and use a cost of $1000/kW inclusive of O&M for a 20 year basis (many utility folks tell me this is a decent assumption). Let’s assume, for now, the cost of gas is equivalent to the LCOE of the sun for solar and electricity into the battery. And finally let’s assume there’s no carbon tax. If we determine the duration of the application against this standard, we can figure out how much a battery can cost for a given duration. Finally, let’s run this calculation assuming the battery lasts 20 years, 10 years, 7 years and 5 years. We get a curve that looks like this.

Figure 1) Maximum cost of battery on power basis compare to DG gas turbines

What this shows us is that storage can be expensive for grid regulation and reliability aspects that are short duration, and moderately priced for peak smoothing (the yellow curve), but it needs to be pretty cheap for diurnal support.

The orange zone to the left is where many grid scale battery storage projects have gone using high rate cells, because the cost of the batteries can be higher. The Powerwall batteries, based on their discharge power specifications, would need to cost $250/kWhr for the daily (7kWhr) battery and $110/kWhr for the backup/weekly (10kWhr) battery. These numbers are lower for the home use case, as a home generator for 2 to 5 hour use is on the order of $200/kW.

So this represents a really hard challenge for batteries. They need to last longer while providing their power more quickly to compete with cheap gas. I was lucky enough to be part of a group of researchers last year studying this in depth, and when we used a more nuanced perspective on gas prices we generated a curve that looked like this:

Figure 2) Lowest Cost for Baseload Power

So without carbon taxes and subsidies and looking at the cost of gas in the last 10 years, it’s difficult to justify the use of the Tesla batteries for solar and wind support as far as I can tell.

It is a different story for transmission and distribution upgrade deferral (TD/DD). Figure 3 doesn’t take distribution into account, Figure 2 does but only implicitly. In NYC the price of electricity $0.22/kWhr. Little over half of that is the cost of distributing the power. Yes, it costs more money to maintain and upgrade wire than it does to generate electricity outside of New York. Power lines degrade faster when taxed. ConEd charges up to 5x to 6x the above price during critical use period to commercial entities. This inclues co-op housing. So these customizers can either curtail use, or artificially “shave” their peaks used either a generator or stored energy in a battery.

I’d argue the yellow portion of the curve still applies to this case and to TD and DD in general. A battery can save someone money, but as far as I can tell a generator can save them more money with current battery technology at the prices above the thresholds listed above.

This is why the High Risk/High Reward arm of the DOE (ARPA-E) is funding people aggressively perusing batteries that can be produced for under $100/kWhr while providing over 1000 cycles. At these rates, even a 5 hour battery can compete against cheap gas and a generator. (Disclosure: I was lucky enough to have received ARPA-E money to tackle aspects of this problem for batteries and power supplies a few times.)

As always, the above is subject to change. Discussion makes everything better.

Show your support

Clapping shows how much you appreciated Dan Steingart’s story.