Electric Grid Myths, Part II: The Effect of Alternatives

In my previous article, I laid out many of the most common misconceptions of how the electrical grid is built, operated, and maintained, and just precisely what is meant by the term “smart grid.” This leads to a number of questions and choices regarding the future -- specifically, how telemetric load control (what “smart grid” really is) works with wind power and electric cars.

Wind power has been heralded by its proponents as the future of electrical power generation. I specifically am not going to discuss the merits of wind power vs. coal, nuclear, solar, or any of the other possibilities. I am decidedly not advocating any particular source. What I am doing is noting the time scheduling characteristics of wind, why that's a problem, and how telemetric load control -- particularly if a significant penetration of electric cars catches on -- might work together to mitigate one of wind's major drawbacks, and why that's probably going to have a limited effect.

Aside from concerns over bird kill, economics, and aesthetics, wind power suffers from a problem with reliability. Not reliability of the machines themselves (though maintenance is a significant operating cost), but the reliability of the wind itself.

A case study: Bonneville Power Administration

The Bonneville Power Administration is a federal agency that operates a series of hydroelectric generators in the Pacific Northwest. Recently, they've added a substantial number of wind generators to the system, and they make the actual operating data available on their website. The data is very high time resolution, and includes aggregate wind production at 5-minute intervals. A sampling of this aggregate power production data at various times of year support BPA's statement that output from wind is “essentially random.” “Essentially random” in this case means that the aggregate output can range anywhere from literally zero to almost two gigawatts, and typically switches between almost zero and over one gigawatt several times a week. It spends a rather large amount of time at close to zero output.

This histogram, created from their online data for the period of January through June 2009, shows the aggregate power output to be up to almost two GW but far more likely to be very close to the low end: with the aggregate output being between 50 and 100 MW about 17% of the time, and between 0 and 50 about 8% of the time, with the remainder of the 50 MW increment “bins” averaging about 2% of the time each. This kind of distribution is rather difficult to compensate for, since it has essentially no baseload output, and requires 100% reserve.

Bonneville, of course, can manage this because they have many times the hydropower capacity as they do wind. Hydropower plants, unlike base load thermal plants, can change output very quickly, and complement wind extremely well -- there's typically a limit on the total amount of water available, and the highly variable wind power allows them to save water for another day.

But this is something that can't be replicated anywhere else in the U.S., and can only be done in a few hydropower-rich locations in the world.

The more typical situation

The more typical situation currently is that there are a few token windmills that are such a small part of the mix that they get lost in the baseload, as natural gas-fired peaking plants, that are there for other reasons, take up the slack. As long as wind is contributing in the low single digits of the percentage of the power, that's not a huge problem, but as it starts to become a major contributor, that becomes a bigger problem.

The problem is somewhat mitigated by spreading the turbines out over a large area, so that the average wind speed at any given time is more uniform. But that doesn't solve the problem. It just makes the need to make up with another source less, on average, and it makes an event of the total wind over the broad area dropping to a very low percentage of capacity less likely. But not impossible.

The BPA data is illuminating, with the entire network operating at between 0 and 2.5% of capacity for 8% of the time, and between 0 and 5% of capacity for 26% of the time.