February 11, 2009

Do wind & solar need storage?

No. It is often argued that wind and solar intermittency create a need for expensive utility electricity storage facilities. I would argue that there several much more cost-effective alternatives: demand side management; geographical dispersion; and using existing generation as backup.

First, covering demand from storage for any significant time would be very, very expensive. Better to handle as much as possible with almost anything else, and use storage as a much lower priority resource.

Second, I believe there is general agreement that wind can achieve a market share of at least 10%, and probably 20%, with current load-following techniques (including modest levels of the alternatives I'll describe below), so wind can grow quite a bit without anything that might seem exotic. If wind captured only 20% of the market, it could displace 40% of coal, or 20% of coal and 50% of natural gas.

1) The first alternative is Demand Side Management (DSM): short term intermittency is far better handled with DSM than with central storage, especially as the number of plug-ins and EV's grows. DSM is almost free to utilities, and has both effectively instant response times and enormous capacity.

Plug-in/EV charging can be scheduled when it's needed. If your problem is too much wind in the middle of the night, charging can go there, and easily be 1/2 of demand. Heck, for short periods it could be as much as you wanted: visualize 150M plug-in's pulling 6KW each, for a total of 900GW!

Plug-in/EV's could also provide V2G, and provide additional supply in similar numbers.
Does it seem hard to imagine that many plug-in/EV's, or hard to imagine them ramping up quickly enough? Well, the thing to keep in mind is that they can grow as quickly as wind and solar: we could easily produce 10M plug-in/EV's per year in 10 years.

We should note that DSM for PHEV/EV's is more important than V2G. It sidesteps battery cost issues, as well as other complexities that come from using wires in two directions. OTOH, it's highly likely that the 2nd generation Li-ion batteries now being put into production will last longer than the vehicles they power, rendering the cost per cycle question unimportant for V2G.

It's important to maintain clarity about the timeframe and context of our discussion. If we're really talking about a grid that has a very large % of renewables, we're either talking about decades in the future, or a world in which our society makes a much, much larger commitment to dealing with energy issues than it has so far. In such a world, a very large number of PHEV/EV's with relatively large batteries is extremely likely. In that case, it's reasonable to assume that we're talking about over 100 million PHEV/EV's, with batteries that can effectively hold 25KHW or more. Such batteries could power vehicles for days between charges, and provide enormous flexibility for DSM (much more than a 8 hour scenario one might consider).

There is enormous potential from creative use of PHEV/EV's, potential that we are far from understanding. I would note just one: the motors in PHEV's are extremely efficient, on the order of diesels. A fleet of PHEV's would provide backup capacity on the order of 500GW that could be sustained for days, using engines that would be as efficient and far cleaner than most diesel generators. Would we want to use such a capability often? Of course not, but it's availability would be enormously valuable.

3) it's easy to exaggerate the intermittency we need to handle: it wouldn't take much interconnectedness to take advantage of geographical dispersion of negatively correlated wind and sources.

4) solar is negatively correlated with wind, both on a daily basis and seasonally.

5) we also have the option of backup by (hopefully) largely obsolete FF generation plants, so DSM (or storage) wouldn't have to handle very long (but rare) events. The US has slightly less than 1,000GW of nameplate capacity. US average generation is about 450GW, so the overall US capacity utilization is less than 50% - that's useful for people to keep in mind: we have lots of extra generating capacity, which would provide a lot of buffer, especially from Natural Gas, which is the most flexible source. That would help make it possible to dramatically expand wind generation.

I'd love to see a really good simulation of these methods. Unfortunately, no one has seen the need, as 10-20% market penetration seemed distant. There have been analyses of the benefits of combining geographically separated wind sites: they found that variance was dramatically reduced, to the point that it seemed reasonable to describe wind as base-load.

Finally, if we insist on storage, it wouldn't cost that much. A kilowatt (nameplate) of wind costs about $2,000. It might need 4 hours of storage at $120 using lead-acid (1KW x 30% capacity factor x 4 x $100/KWH) - that's not so much.

Here's an article about Google's effort to facilitate such things with in-home power monitoring (hat tip to Bob G).

No comments: