Wind power: debunking geographic smoothing

John Petersen at Seeking Alpha has being doing his homework on the wind-advocates’ claims that intermittency is easily solved by spreading wind farms over more geography. Like David MacKay, John examined the numbers. In this case the real-world wind power outputs reported literally across the planet, “over 17 timezones and two hemispheres”.

If you are skeptical, you can invest a few hours on Excel to test John’s analysis for yourself. John downloaded the public data from five power authorities in Pacific Northwest,

Australia, Eastern Canada, Western Canada, and Ireland. Then he synthesized a global electric grid connecting these five regions into a single power supply. You really can’t ask for a more perfect test of the validity of “geographic smoothing”. These following two sample graphics illustrate the reality — that even with this hopelessly optimistic global grid, there will be periods where the wind output is tiny – periods that are too long to compensate with any affordable energy storage option.

Please do read the whole thing. For another in-depth look at geographical smoothing, please see the two BNC guest posts by Finnish physicist Jani-Petri Martikainen, discussed in my earlier post “High renewables penetration means eye-watering costs and massive overbuilding“:

There are two posts, intended to be read in the following order, where the first analysis examines unreliable wind added to an existing reliable grid, the second analysis adds solar to the mix:

[1] Geographical wind smoothing, supergrids and energy storage

[2] Solar combined with wind power: a way to get rid of fossil fuels?

Based upon real-world data sets, with the data chosen to be extremely favorable to wind and solar, the simulations indicate that an idealized case of optimal solar/wind deployment requires a massive overbuilding of dispatchable power sources.

3 thoughts on “Wind power: debunking geographic smoothing

  1. Actually, it should not be necessary to prove that area smoothing is incapable of making wind power reliable. It is the responsibility of those making the claim to prove it.

    Let us suppose that it were possible. With an availability factor of 20%, that means that wind farms would have to have five times the capacity that a fossil fueled or nuclear plant would require; that would greatly increase costs. Then too, the power would have to be gathered from widely dispersed locations thereby adding even more to the cost. The excessive costs would probably be unacceptable politically.

    It would be interesting to calculate the actual costs.

  2. I would assert that this should not be regarded as a matter of overbuilding, but rather underestimating capital costs for reliable operation.

    But the “reliable” side you cite has its flaws, too. Consuming irreplaceable fuels that are also needed for transport, or carry substantial pollution costs and risks. Large plants that fail or are shut down for maintenance with a large impact on the overall grid. The risk of fuel supply disruption — labor strikes, transportation disruptions (for example, floods), mining accidents. The risk of price shocks — and their upward pressures on fuel prices in general — effectively costs that end up being counted elsewhere. Large facilities as targets for terrorist attack.

    If we “overbuild” solar and wind and similar technologies, and overbuild hydro’s peak capacity and build retaining ponds to even out the flow downstream, we could trade high capital costs for low operating costs and low pollution. I don’t know how that would cost out — especially upgrading hydro peak capacity. I’m just suggesting that broadening the portfolio of sources further helps to reduce the problem.

    If we consider the renewable sources primary and secondary, and fuel-burning sources as tertiary, the question is how much “overbuilding” of fuel-burning sources do we need to make a diversified system work? How much fuel are we willing to burn? How much risk of shortage are we willing to incur? How much price risk?

    Is it cheaper to overbuild or build storage capacity? Storage is more predictable, but expensive. But giving the same degree of reliability with overcapacity is more expensive as you demand higher statistical reliability.

    This can’t be reduced to a one-dimensional optimization problem, although one-dimensional analysis can aid understanding of particular trade-offs.

    Finally, while I applaud attacking the question with Excel spreadsheets, using just five sites is a statistically flawed analysis, regardless of how geographically diverse those five sites are. Unless, of course, you want to test whether five would be enough diversity. It certainly illustrates that the problem cannot be solved with a small set of sites, but does nothing to show that the problem cannot be solved with diversity of sites.

    I don’t think it can be solved entirely that way — but we would need to analyze many more sites. I’d suggest doubling it to 10 sites, observe the effect, and double it again to 20, until you can get an empirical model of the trade-off between diversity and stability of supply.

    I could suggest more theoretical approaches, but given that real-world sites aren’t likely to be completely independent in their conditions, I think a more pragmatic approach is better.

    That’s a lot more work than an afternoon with a spreadsheet, but unfortunately, I think that work is needed to get a real picture.

    It certainly needs to be analyzed, and not just assumed, and even a simplistic analysis is sufficient to demonstrate that.

    I guess I should go post on John Peterson’s site and encourage him to explore the question further.

  3. Bob Kerns, I especially like the following from your post:

    “That’s a lot more work than an afternoon with a spreadsheet, but unfortunately, I think that work is needed to get a real picture.

    “It certainly needs to be analyzed, and not just assumed, and even a simplistic analysis is sufficient to demonstrate that.”

    Many readers will be familiar with one of the definitions of assume. Although it is very slightly off-color, I will state it here with the hope that I will not be banned for doing so:

    ass u me, i.e., assuming makes an ass of you and me. I recently experienced the results of assuming when the telephone company incorrectly assumed that my telephone problem was the result of a damaged cable in the area and therefore did nothing until I threatened a lawsuit. Assuming can be very risky.

    One of the things that concerns and puzzles me is that many environmentalists are urging us to create an energy system, at a cost of untold billions of dollars, for which an adequate analysis has not been done. I would think that anyone with a good background in economics, accounting, and physics, would see the folly in such action. It seems to me that those subjects should be required to get a degree in anything and that we are seeing the results of failure to do so.

Comments are closed.