How predictable is wind power?

I have sometimes heard people say wind power is reliably predictable, but have never looked into the issue before, mainly because some of the people I have heard make the claim are not likely to be stupid enough to have this wrong.

It’s relatively easy to check the issue courtesy of the Irish electricity grid. Ireland has both a relatively high wind penetration (greater than 11% of electricity) and freely available data on forecasted and actual wind generation. Below are screenshots from their website in the last three days:

This looks a pretty decent fit. What about the entire year so far? Comparing actual with forecasted wind generation shows a very strong fit (statistics note: correlation coefficient of 0.95 and R^2 value of 0.9 in a linear model.) Looking at individual time periods back this up further. So, it looks as if wind power is reliably predictable (at least in Ireland).

About these ads

5 thoughts on “How predictable is wind power?”

  1. I have looked into similar things some time ago. I suspect that it is not this simple. I wrote about this into my own blog. Posts are in Finnish and translate into English is crappy, but maybe you can figure out the most relevant points: http://passiiviidentiteetti.wordpress.com/2011/01/18/tuulivoimavuosi-2010-tanska-ja-irlanti/ and http://passiiviidentiteetti.wordpress.com/2010/03/23/tuulivoima-case-study-bonneville-power-administration/. The prediction error naturally depends strongly on the time period so that short term forecasts are better than longer ones. However, the prediction error is not normally distributed and very bad predictions do occur at a much higher frequency than what you would like. E-On Netz has some years ago written about their experiences in Germany (wind report 2005) and prediction errors that were a substantial fraction if installed capacity (and comparable to average output) did happen.

  2. Thanks.
    I had a look. It’s probably true that there are occasions when the predictions are out a lot. However, it’s unclear if this is going to cause any real problems for the grid. The grid will already have to handle wind producing essentially zero electricity on very low wind days. Presumably if the grid can handle this, then it can handle the forecast being very incorrect occasionally.

    1. Yes, for small penetrations especially the system can deal with it. If one aims at higher penetrations problems probably mount. If one accepts nuclear, as I believe we both do, then it is harder to see what value the additional wind capacity gives. Maybe it is required for political reasons, but I still haven’t seen an explanation why it makes technically more sense than pure mixture of nuclear and gas in the short term and pure nuclear in the longer term. (If some miraculous revolution happens in storage techs AND renewable costs things could be different, but I do not put too much faith in miracles riding to the rescue .)

      1. Wind can make some sense together with hydro-power. Hydro-power is about the most efficient to follow load as it’s efficiency does not depend on the load.
        The recent CCGTs from Siemens, GE and Alstom are making progress, but the efficiency still does go down at lower loads. And also there’s a load threshold below which it’s impossible to use a CCGT because you don’t respect the pollution levels anymore (this varies between 60% for older CCGT and 15/20% as the absolute record claimed in white papers). So when load gets to low, it’s best to stop it and restart later when needed. But even if frequent restarts are possible with the newest generation, they still burn much gas for nothing. So you need to stop for at the very least 3 or 4 hours to make this efficient (this is where it’s most useful to have a correct wind prediction). All in one, whilst the very recent CCGT model are progressively making it easier to follow load, it’s not the case with older models, and it’s not perfect even with them.

        Also when you use wind together with gas, you are locking in that gas. you use less than you would in a pure gas system, but it will be very hard economically to get rid of that part. Any replacement will have to share the economical space with the wind. Since you already get penetrations problems with wind at a level where it’s still not a huge part of your mix, the gas complement is a *lot* of gas. Spain and Portugal are the two countries where you can see that effect most. They both strain their grid to the limit from time to time with the wind production, but this happens with a wind level that’s IIRC over the year only around 20% of the mix. Danemark on it’s side has the easy option of exporting to hydro-abundant Sweden, even so it’s not very great economically.
        Whilst with hydro, you lock in that hydro, which is not a CO2 problem, and what’s best is the partial production is not even an economical problem for the hydro, because the water availability constraint means it can’t anyway run at full load all year round. The Portuguese have a lot of hydro planned currently, in order to easier things obviously.

        However whilst hydro+wind can be made to work, and be efficient for strong CO2 reduction, it’s still a rather expensive solution, especially when you don’t forget to include the cost of grid reinforcements directly caused by wind.

  3. Thanks Jean-Marc. You make good points. Of course hydro is a limited resource in most countries. In Europe I guess most of the practical potential is already in use and its share is …what perhaps 10% ? This limits the role wind+hydro combination can play. Some countries have a higher share, but most Europeans live in pretty flat or dry places.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s