SWEET: A Wind Energy Prediction Model

[Mel Tyree]

I have had some time to take a closer look at the Cadmus Group report submitted to the MTC, i.e., the agency that provides incentive $ for small wind turbines in MA. SEE:

http://www.ualberta.ca/~mtyree/SWIEP/Docs/CadmusGroupReport20041408.pdf

The Cadmus Group reviewed a large number of installations funded by MTC and eliminated the units with obvious technical problems or with too little production data then produced Fig. 5, which gives actual annual production as a fraction of the amount predicted by the installer. So a ratio of 1 means the turbine produced as many kWh as predicted. Although prediction models are unlikely to be accurate for any one turbine & corresponding site, we would expect the mean of the ratio of a large number of sites to = 1, i.e., not be significantly different from one. No stats are given in the report so I computed the mean, standard deviation and it equals 0.245 +/- 0.127 N = 19. The probability that this mean is NOT significantly different from 1 based on random errors is 1 in 10^15 (one in 1 million billion). Hence there are problems in the industry with the ability of models to predict real-world results.

Most of the 19 turbines in this study were Bergey 10 kW models, and hence the predictive model used is much like those used routinely by Rob Beckers and Bergey Inc., i.e., likely to be state-of-the-art models.

Cadmus has an improved model (SWEET) that gets the estimates closer to reality, but the stats still aren’t great. I will upload this model and ask that interested parties take a look at it. The major difference is that Weibull wind probabilities are given for different compass directions in a wind-rose and turbulence is estimated in each wind-rose direction. I would like to have comments from anyone willing to study the model.

I am still digesting it myself.

[Rob Beckers]

Interesting report! Thank you for posting it Mel.
I've often wondered how the average wind turbine install fared in comparison to predicted production. Installers don't like to share this data (if they even have it). According to this report the real-world installation results in terms of production are not good, to put it mildly. With a range of less than 10% of predicted for the worst, and only about 60% of predicted production for the best.

Mel, do you know how "average capacity factor" is defined in this report? I can't find that in there.

I have to digest a little about what was said in the article. All the reasons they mention make sense. From personal observation, I've noticed a tendency of installers to put wind turbines on short towers. The rule of "30 feet above any obstacle in a 300 feet radius" gets violated more than a little, and that rule is actually an absolutely bare minimum that still results in significant turbulence at hub height. From looking at efficiency numbers (easily calculated in a spreadsheet) I know that there are lies, damned lies, and manufacturer's turbine power curves. Many of the published power curves would require unrealistic efficiency numbers for the turbine, indicating that generous liberty was taken in creating that curve. Some manufacturers are worse than others. Installers also don't want to be too 'realistic', since customers will not buy if they know how little their turbine is actually going to produce. Of course, this strategy only works in the short term and gives small wind a bad name.

When I have a bit of time I'll check out the SWEET model. Sounds like it would be a way to improve accuracy. Unfortunately I'll be gone on a trip the next 4 weeks, so that'll have to wait.

By the way, I was happy to see that the one Scirocco wind turbine listed in the report came out at the top! :)

-RoB-

[Mel Tyree]

Mel, do you know how "average capacity factor" is defined in this report? I can't find that in there.

-RoB-

This is maybe one of the dumbest things about the report. I think what they have done is take the 'nominal' rating of a turbine x number of hr in a year and divide that into the actual production (kWh in a year). I am not sure if this measure is the idea of MTC or Cadmus!

Don't shoot me, I am just the messenger!

[Steven Fahey]

Hi,
I just joined, but I've been interested in wind power systems for a while. I read the Cadmus report, and the rebuttal of that report by Paul Gay, too. Rob Becker's concerns below reflect that, too.

I started wondering why their analysis didn't just give the site roughness factors - wouldn't they reveal a lot about the capacity factors? A chart showing R versus CF might reveal something. There isn't much worth reading in their analysis of the "shortfall", that people who install this stuff don't already know.

In the past, I have found technical papers written by think-tanks and consultants that found "problems", and offered solutions. Solutions they are all too happy to provide. It's all good business and fun to be competitive, but don't trust the technical info.

I didn't start worrying about that until they started nitpicking the Wind Atlas. Are they angling to do the number crunching on a new national wind energy survey?

[Mel Tyree]

Hi,
I started wondering why their analysis didn't just give the site roughness factors - wouldn't they reveal a lot about the capacity factors? A chart showing R versus CF might reveal something. There isn't much worth reading in their analysis of the "shortfall", that people who install this stuff don't already know.


Can you give me any instructions/references on how to measure site roughness and is there any way to predict turbulence intensities from site roughness at any given wind velocity?

See page 13 of the attached file if you are not sure what I mean.

[Steven Fahey]

Here's what I referred to when looking at my own site:

http://www.windpower.org/en/tour/wres/shear.htm

and,

http://www.windpower.org/en/stat/unitsw.htm#roughness

The formula is rather simple to work with, but picking the correct factor for your site can feel like "throwing a dart". I've done a "best guess" at my site, and it will be interesting to see how moving my anemometer around will affect the apparent wind speed.

Hope that helps.

(what attachment? Sorry if it's there, but I don't see it in the body of your message.)

[Mel Tyree]

Rob,
I found it defined! Look at page 9-10 of the pdf file in the previous message from me in this message stream.
CF = 0.087Uhub -Prated/D^2
where Uhub = the mean annual wind speed at the hub height in m/s, Prated = rated power output of the turbine kW, D = rotor diameter in m. Now this equation make dimensional non-sense unless some units are included in the 0.087 because I make out Prated/D^2 to have units of m kg s^-3. Perhaps the equation is misprinted.

I guess I need to read the book it came from G.M. Masters, Renewable and Efficient Electric Power Systems, Wiley, 2004.

To see the full report go to: http://www.ceere.org/rerl/rerl_resourcedata.html where you can download a large number of wind resourse reports in MA for FREE!
--Mel

[Mel Tyree]

(what attachment? Sorry if it's there, but I don't see it in the body of your message.)

The file was too big to load and I didn't notice. It is one of many files at
http://www.ceere.org/rerl/rerl_resourcedata.html

Thanks for the Danish wind industry site. I have seen it before but never fully explored the site. It obviously has a wealth of useful facts and data.

[Steven Fahey]

Okay now the report is visible...
This will be interesting to read.
Thanks.