The answer is blowin’ in the wind

Recently I discovered a story about wind farms that suggested that there was a “…possibility that hundreds of derelict wind turbines, coated with indestructible toxic chemicals, will be left rotting on Welsh mountains for hundreds of years.” The basis of this was a report from the Renewable Energy Foundation (REF) that claimed that the “normalised” load factor of wind turbines declines dramatically over time.

[The load factor of a wind turbine is defined as the amount of energy produced in a given time by the turbine, divided by the installed capacity of the turbine multiplied by the amount of time it was generating energy for.]

If the load factor of wind turbines is decreasing as they age, the implication is that wind turbines are generating less energy as they age, and therefore they are a bad investment. We shouldn’t be wasting our money on them, and so on and so forth…In fact they state in their report that “The results of the statistical analysis demonstrate an unambiguous and statistically significant decline in the operating performance of wind farms as they grow older.”

Helpfully, the report includes a reasonable description of its methodology, and the data they used is available to download from their site. This is where it gets weird(er). The report includes a box plot (Fig. 8) of load factor by age for UK onshore wind farms and there is… no clear trend in the median value of load factor by age.¬† If you plot wind farm output against time for any of the wind farms they have in the data set it is easy to see why.

The output of the wind turbines doesn’t really change over time. It goes up, it goes down, but over time it doesn’t really change too much at all. For Example, Figure 1 shows the energy produced between April 2002 and February 2012 by the Four Burrows wind farm. Four Burrows is a 4.5 MW wind farm which came on line in 1995. The farm is made up of 15 x 300 kW Siemens wind turbines and is situated near Truro. It is fairly typical of ’90s wind farms in the UK. You don’t have to take my word for it, you can download the data and see for yourself with any of the wind farms in the database.

Figure 1: Four Burrows wind farm output.

The question is then, how do they come to the conclusion that load factors are decreasing over time? Well apparently “…the distributions mask crucial differences between the performance of wind farms as they age because they do not control for the differences in location and wind availability over time.”

This doesn’t actually make a lot of sense. If we remember the definition of load factor, we can see that all it does is tell us what proportion of the installed capacity of the wind farm is generated over a given time period. If we consider that the energy produced by a wind turbine is dependent on the wind speed at a particular location and time, then load factor would appear to account for differences in wind availability over time.

Regardless, they go on and produce a fixed effect model to take into account the “previously unaccounted for” locational and wind speed effects and produce their answer. Now, I am not a statistician, but my suspicion is that this model is not suitable here, or as above, its not really even necessary to use it for the purpose the report puts it to. Luckily, my thoughts on it are not important, because it doesn’t actually make any meaningful difference to the wind industry if this method of analysis is correct or not.

What matters is the amount of energy actually produced by the wind farms. Even if you agree that the fixed effect model is a reasonable way to determine a “normalised” load factor for wind farms, and that this “normalised” load factor does decline over time, it doesn’t matter. The same data¬† shows that the amount of energy produced by those wind farms is not dramatically declining over time, and so the performance of the wind farms is also not dramatically declining over time, contrary to what the report would like you to believe.

So why would you go to all that trouble to produce a “normalised” load factor metric? The answer is blowin’ in the wind…



This entry was posted in Statistics, Wind Power. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *