windA while back I claimed that an all-dispatchable grid would require 75 hours of storage. I have been definitively proven wrong. Nothing like that is required.

In fact, tiny amounts of storage may do the trick if the grid is large enough — and making the grid large enough is not that expensive. We may already have in place what we need for a completely renewable grid.

The 75-hour figure came from studies of single, isolated wind farms. But as you add wind farms, the odds of two wind farms being down — or low producers at the same time — drops.

I came across confirmation by accident: a Vehicle to Grid study (PDF) that evaluated data from eight sites showed that storing just 36 minutes of nameplate capacity would allow the widely dispersed farms to meet a firm power commitment ~90% of the time.

There are tricky steps involved. I am not necessarily advocating a 95%+ wind grid, as there are many ways to generate renewable electricity. But just as coal dominates power production in our current system, I suspect wind will dominate in any future grid. So consider the following a limiting case for wind.

Wind generators hit maximum speed only rarely — on average, they generate about one-third of nameplate capacity. (Output is spread out over much more than eight hours a day, though.) This is not a problem; capital and operating costs are still low enough to make their power competitive with fossil fuels.

However, were the wind industry committed to firm delivery, it would be at some percentage of nameplate capacity — probably a percentage well below one third. 20% is the figure used in the study cited.

20% of peak would be available for each of the eight sites studied around 80% of the time, even without storage. Measurements from the sites showed that production dropped below that level for three hours or less another 10 or so percent of the time. The stored equivalent of three hours of 20% of nameplate rating would equal only 36 minutes of that nameplate capacity — and meet a commitment 90% of the time.

The important thing about the study is that a result for eight sites is going to be pessimistic, not optimistic. If wind is to provide a substantial percentage of our power, it is going to come from hundreds or thousands of widely dispersed sites — which every study agrees will reduce the number of hours of low power (example, example, example, all PDF).

Another constraint is widely agreed upon. Yes, wind can easily meet a firm commitment at 90% availability with only three hours storage. But given how much wind power is produced at night, when demand is low, is three hours enough storage to prevent having to dump much of the remaining power?

The answer is, we don’t know. We do know that dispersion of wind farms reduces difference between minimum and maximum production. With a thousand farms, you will never see production rise anywhere near nameplate capacity, and seldom see it drop below 10% of rated capacity. So three hours storage might well be enough for overnight demand lulls as well.

What if it is not? In the absence of hard information, let’s assume we install ten hours storage. That’s enough to substantially time shift supply, especially since there will never be a need to store 100% of output — some percentage will always be used as generated. Storing ten hours of a firm 20% commitment represents just two hours of peak capacity.

So even with these unrealistically pessimistic assumptions, two hours of storage (relative to wind farm nameplate capacity) will allow the grid to be 90% powered by wind! (As it happens, this much storage would allow the grid to be 95% rather 90% wind-powered — only five percent of low-wind hours occur in blocks of more than ten hours, even with only eight sites.)

Three questions remain. What kind of storage is available? What do we use for the other 5%? And what would electricity based upon this model cost?

The cheapest storage is pumped storage. There are questions about the carbon neutrality of hydroelectric dams, but they seem to apply more to low-head hydro than high-head. If even the worst fears prove true, dams 700 feet or more above the low point they dump to will probably still be usable.

Pumped storage can add as little as 1 mill (.1 cents) to the cost of a kWh. But because we are asking for more storage capacity (per KW of generating capability) than usually demanded of pumped storage, the price would probably range from 2 mills for a three-hour buffer to around 7 mills for a ten-hour buffer. Currently, we have enough pumped storage to store about 19.5 gigawatt hours — around what the U.S. consumes every 25 minutes.

If pumped storage does not work, we could use Vanadium flow batteries — more expensive, but as we shall see, tolerably so.

What about total generation and additional transmission cost? Wind power can be generated at 4 cents a kWh unsubsidized (only on certain sites, but if we are going to have a national grid using HVDC lines in any case, we may as well be fussy about where we place wind farms). Also, with this large a usage of generators, we can demand wholesale prices. We can get the best financing. As we increase the size of wind farms, we can gain economies of scale in site preparation, permitting, management, operations, and maintenance.

Because HVDC lines can be amortized over long periods, as long as 40 or 50 years, their cost is extremely low — perhaps 1 cent per kWh. Transmission losses will add another 10% (4 mills per kWh).

Combine this with pumped storage and the total price for a 95% wind mix runs from 5.6 cents per kWh for 3 hours to 6.1 cents per kWh for ten hours.

If we used flow batteries ($350 per kWh of capacity, 30% round trip storage losses, 1 mil O&M), the added costs would bring the price of wind up to around 6.8 cents per kWh for three hours of storage, 8.6 cents total per kWh for a ten-hour buffer.

Again, there are renewables other than wind. But just as a limiting case, it is interesting that we could provide 95% of our power from wind — reliably — for between 5.6 and 8.6 cents per kWh. Note that existing hydroelectric could provide about 80% of the remaining 5 percent, with biofuels or natural gas-powered generators providing the remaining 1 percent. Alternatively, we could provide 5% of our electric needs with sustainable biofuels. (Whatever arguments we have about biofuels on a large scale, does anyone doubt we could produce enough sustainably to power 5% of our electricity needs?)

Although wind would provide 95% of our electricity, there is also the question of operating reserves. How much energy (as opposed to power) do we need, how much generating capacity?

Again, this is tough to answer — it has not been studied enough. Some studies suggest you’ll never see generation on a large scale wind grid drop to more than 10% of nameplate capacity for more than a few hours. If true, that would mean a near-100% wind grid would require operating (not spinning) reserves of 50% total peak grid demand.

But let’s be unrealistically pessimistic and assume that occasionally wind would drop to the point where it produced no output anywhere in a 2,000 mile diameter wind grid for more than ten hours at a time.

In this highly unlikely case, existing fossil-fuel power plants could run on biofuels, as long as they didn’t get run often enough to burn an unrealistically large quantity of such fuels. Coal plants could be converted to run on charcoal. Natural gas plants could run on syngas. Peaking diesel plants could be run on biodiesel. In some cases, they could run without modification. In short: operating reserves would cost whatever was needed to keep existing plants operational and run them occasionally on biofuels.

Capital and O&M costs of conversion and maintenance are difficult to estimate. Again taking an absurdly pessimistic assumption, let’s assume 1 cent per kWh over the entire production of the grid. (This is the equivalent of spending 180 per KW of capacity for conversion, and then spending another mill per kWh for O&M.)

Update [2007-1-16 11:45:51 by Gar Lipow]: In practice, while we have existing peaking infrastructure than can respond fairly quickly to outages, many of our existing plants are not designed to be brought up quickly from a cold start, and keeping them “warm” consumes energy and adds maintenance. To the extent we could not use existing infrastructure, standby diesel costs betwee $200 and $300 per KW. Given that we are talking about using them less 100 hours per year, neither relative thermal inefficiency, nor lifecycles as low as 1,200 hours prevent their being an economical choice – still adding less than a cent per kWh to the total grid including biofuel and maintenance.[/UPDATE]

Lastly, there is the cost of the biofuel itself. Biofuel to provide 5% of electricity, with costs spread over the entire output of the grid, is unlikely to be more than another 2 mills. If hydroelectric was providing most of this backup, the cost of biofuels would be less than .4 mills.

So total costs range from less than 6 cents per kWh to around 10 cents per kWh hour for a 100% renewable grid, 95% powered by wind. The pessimistic end of this spectrum requires more extreme assumptions than the optimistic end — I suspect actual costs would be nearer to the low than the high end.

Again, the whole assumption of 95% wind is itself unrealistic. I doubt we will ever want such a “monoculture” grid. A real renewable grid would run from a mixture of sources.

But it is more than a useful limiting case. At minimum it implies that wind (mixed with a small amount of something else for long outages) would be useful for something no one thinks wind is good for: base load. Suppose you had wind farms set up to provide a “firm” commitment for base load with 90% availability.

Typically base load is about 40% of total consumption in a grid, though this varies widely. As we have already seen, three-hours storage provides this with better than 90% availability, even with a small number of sites. By using this for base load you guarantee the “firm” commitment would be useful. You would never have demand for less than 20% capacity, which would be equal to that base load. Further, much of the power generated above the 20% level would occur during times when demand was well above base. Possibly the three hours by itself would be enough to time shift additional power to when it is needed. If not, no more than hour or two more is needed for such time shifting. So total storage requirements would be four or five hours of whatever percentage of capacity was contracted as a firm commitment.

Increasing storage to the full ten hours (or whatever would increase availability to 95%), it might well make sense to let wind provide more than base — to provide as much as 65% of the electric grid. While in practice we may never want anything close to an all-wind grid, we may well find that a wind/hydro dominated grid makes sense.

If you were to put in place today the cheapest low-carbon grid, it would come close to the 95% wind figure I mentioned. You would use existing hydro, new and existing geothermal, solar thermal in the deserts, and photovoltaics on tall buildings and other places where they can replace expensive structural material and also in sunny climates where concentrators could be used. You would take advantage of wave and tidal to some extent, along with biomass. You would use small quantities of natural gas, because fossil fuel use only needs to be near zero.

With all that, you’d still end up with somewhere between 80% and 90% of power supplied by wind — the lowest priced renewable alternative in the U.S.

Of course this is not something you would do all at once. As you phased in such a wind-dominated grid, the price of other renewables would drop. So in practice, if we were to use wind plus long-distance transmission and storage to begin decarbonizing the grid today, the new capabilities and increased purchase of variable-source electricity would encourage more investment in other renewable electricity sources (such as PV). Wind would probably come into balance with other renewables at the point where it provided 45% to 65% of electricity.