We all know wind power
is intermittent. I thought I would take a look at how predictable
windpower intermittency is and how accurate wind power output
predictions are.
In the process I think
I have tripped over a new issue regarding wind generation (but more on
that later).
The bmreports site
(HERE) has a section on wind generation and shows a graph of an
original output prediction, a more recent and accurate revised prediction and
also the actual out-turn.
The original forecast
value is done 41 hours before the start of the forecasted day. The
revised forecast is done 4 hours before the start of the forecasted
day. The predictions use weather forecast data for the known wind
farm locations and factor in a range of other parameters. These are
sophisticated predictions and are probably as good as they get.
Here is a few example snap
shots taken in the last month from the bmreports site.
But as well as this
continuously updated graph, bmreports also publish the same 3 day
data as an xml file. I've collected the xml files for one month. (Annoyingly I missed two days so these have been left out. But even so
I think this is quite a good data set.)
Above is a graph of
this data showing the final revised prediction done 4 hours before the start of the predicted day (red) and the final out-turn (blue).
(I've left out the original prediction as for obvious reasons, it was more in error of
the final out-turn and so added little to the graph). Note: Actual metered capacity is actually 8972MW so the graph is unduly kind stopping at 8000.)
The first thing that
can be seen is that the out-turn is often (not occasionally) in
significant variance with the predictions. The graphs may be the same shape but the values at any one point in time are often significantly different. Clearly, any system with a
large wind component that relied completely on even a near term
forecast (and without spinning reserve) would soon end up in deep
trouble.
So whatever the
pro-wind zealots preach on Twitter, the problem with lack of wind
power predictability has not gone away. Neither has its
intermittency.
But potentially the
example bmreports graphs (as well as mine) also show another problem.
Notice in the above
graphs how on the occasions the wind output rises above about 30%
capacity (3000MW), during the rise, the out-turn lags the predictions and the maximum out-turn
is significantly less than either of the predictions.
It is as if a large proportion of wind turbines exposed to a rising wind and high wind periods are
being feathered (or throttled back) for some reason during these periods.
At lower wind speeds
there are still periods of great discrepancy between prediction and
output but the tracking between prediction and output does appear
more coherent. (Remember this forecast was done 4 hours before start of predicted day!)
Wind energy companies
only get paid when they are generating. So why would they throttle
back their turbines in high winds?
And the answer to that
I believe is good old repair and maintenance.
For almost any machine,
if you run it lightly it lasts longer. Take a car. The harder you
drive a car the more wear and tear it suffers. Just about all
rotating machinery obey this simple rule – including wind turbines.
We know that there is a severe generic problem with wind turbine gearbox reliability. (See
This Post - The Ghost in the Gearbox and
Post - More Ghosts in the Gearbox )
I would suspect that it has been found that if
the loading on a wind turbine gets above a certain value the wear rate and maintenance/repair cost will be far more than the
return from the extra energy generation.
So maybe operators are unilaterally and quietly deciding that when the wind gets too changeable or too strong, the turbines will be run at reduced output compared to what they are supposed to be capable of.
Wind turbines are
capital intensive. If you suffer a catastrophic failure you will ruin the huge and guaranteed profit (subsidy included) your turbines can make. Do it too many times and you may end up going bust. Better to ignore the whole reasoning, propaganda and hype associated with why the thing was built in the first place and go for the low hanging fruit.
Remember most windfarms
have a 25 year subsidy regime locked in place. It's a nice little earner. Operators are going to
do whatever it takes to maximise the financial gain over this period
and if that includes reducing output to make their gearboxes and other expensive components last a bit longer then they will do it. The abatement of Carbon Dioxide can go to hell.
This, of course, makes an even bigger mockery of the often hyped “Installed Capacity” figure than it already is. It also shows how the unreliability of these machines impacts the supposed reason they were built.
It means that wind power is perhaps even more useless and
under-achieving than first thought.
I cannot prove
the operators are intentionally throttling back their turbines to reduce their maintenance bills.
But I bet I'm right.
