The Unpredictability Of Wind Power
By Paul Homewood
h/t Stuart Brown
https://www2.bmreports.com/bmrs/?q=generation/windforcast/out-turn
It is claimed that although wind power is variable, it is also predictable. After all, we are assured, weather forecasts are now so accurate that they can predict wind speeds.
It turns out that this not the case. Above is the chart from BMRS, showing the initial wind power forecast with the actual outturn over a two-day period.
Below is the explanation of the National Grid forecast:
Based on historical outturn data and detailed local wind forecasts, National Grid forecasts likely levels of wind generation for windfarms visible to National Grid, i.e. those that have operational metering and that are included in the latest forecast process. The forecasts are produced for the period from 21:00 on the current day (D) to 21:00 D+2.Wind Generation forecasts are produced by National Grid’s own second generation windpower forecasting tool. The predictability of the wind varies with atmospheric conditions and so there may be periods where National Grid’s forecast and outturn values differ significantly. Please note that the downloadable data will contain gaps for Original and Updated Forecast values in Settlement Periods that National Grid do not provide forecast values for.
Even yesterday, we can see that wind output was overestimated by more than 3GW, or a third of the forecast.
The “Latest Forecast”, which presumably is made a day before, is naturally more accurate, but has still been up to 2GW adrift.
Fortunately we have plenty of CCGT capacity around to fill these gaps. But when wind capacity has been tripled or quadrupled, such we could be looking at shortfalls of 10GW or more.
Comments are closed.
It isa very common belief that weather forevasting is more reliable from looking at the heavens unassisted b a supercomputer, so to predict windspeed from the day before is nomore accurate than climate alarmists’, inclding Greta’ s, predictions on climate changes.One never gets acknowledgement from meteorologists, nor climate catastrophists that they boobed!
” Lying” is the word, near enough!
Respected countryman Robin Page noted in a recent book that as technology has increased so the accuracy of weather forecasting has decreased. There are plenty of areas where the much vaunted AI can’t compete with human experience. This has been shown up with driverless cars and when the conditions are really bad, who lands the aeroplane – not the auto system as it can’t cope. More weather forecasts – if not all – come from the computer. in days gone by, certainly at the BBC the on screen presenter, being a MetO qualified person, would assemble the forecast interpreting the data.
Ah. Driverless cars. Here’s the test I want to see explained: Say that four driverless cars are running in line on a fairly clear country road. The lead car, for whatever reason is running slow. Which of the following three will take the decision to overtake, in what order, and will it be one at a time? Or do they all slow down and stay in line? (I assume that 5G/AI will allow them to know that they are all driverless – but then again….).
Another test for driverless cars:
Motorway conditions. Fields either side. Cows on one side. Because of a gale the night before (might as well be topical), the fence has blown down, and the cows get on to the carriage way.
What do the driverless cars do?
moo the way out…..
Paul, The reason why the wind is unpredictable, is because it is the result of a non-linear physical system, where the variance is power-law, not linear variance based on a closed population of data. Both weather and climate are subject to non-linear laws and the application of such ‘closed’ population statistical concepts as ‘standard deviation’, ‘variance’ etc to both weather and climate data is a nonsense. I suggest you read, ‘The Misbehaviour of Markets’ by Benoit Mandelbrot, where the reasons why the application of linear statistics to the behaviour of physical systems has never given correct long term predictions are made very clear. The bell curve and all the other mathematical fantasies have strict mathematical restrictions of application. such as complete independance of all points in the data set, which is never true of ‘real’ system dynamics. The computer modelling used by the climate fanatics is therefore incapable of correctly predicting trends or outcomes, unless it takes the non-linear nature of the system dynamics into account – while Gauss fully understood the limitations of the mathematical model he generated, his insight on limitations is not taught in universities today – only his algorithms!
If non-linear statistical ideas were applied to the data you have so painstakingly investigated, then the more riduculous predictions would be found to be false and groundless. Currently, there is little sign of that occurring and both sides are engaging in mud-slinging to no good effect. Claims that the ‘science’ is irrefutable are nonsense when it is clear that the science is based on the application of the wrong mathematics!
In English: computer models are created by programming behaviors. The model run is started by inputting values.
Climate models fail because our knowledge of weather behavior is limited. You can’t program what you don’t know. Like cloud behavior.
Climate models fail because our knowledge of weather data is limited. You can’t get a valid result from a model without valid input data. Which we don’t have.
“…the wrong mathematics…”
Wrong because simply contrary to observation:
“…the wind goeth by streams and not all together…”
Roger Ascham, 16the Century:
ftp://ftp.library.noaa.gov/docs.lib/htdocs/rescue/whytheweather/1927/19271209.pdf
I am hopeful that Boris Johnson is setting up Gove to take the rap for faulty climate advice, prior to reviewing and u-turning on climate policy.
I cannot imagine that Gove’s treatment of Boris in the past is water under the bridge, and the PM is building him up so that the fall is that much bigger.
Not a chance.
As petty as our politicians are, Borisconi won’t be breaking step with the rest of the Western World.
Even Trump’s not been able, or willing, to do that.
The huge amounts of money, investments & taxes, that rely on the continuing climate scam, mean that.
Look at the way that Climategate was brushed under the carpet, it’s completely forgotten now, as far as the media’s concerned.
I do hope there is an exit plan for the Zero Emissions suicide pact. People do not realise that industries are already planning to wind down investment if it goes ahead. It is not just how long a new car will last.
Alok Sharma, the new BEIS minister, is the COP 26 president, in Glasgow, in November.
inews.co.uk has reported:
“In a speech last month at a UK-Africa investment summit, Mr Sharma spoke of the “undeniable implications” of climate change.
“The huge burden of climate change will not be shared equally or fairly across the world,” he added. “Many developing countries are already bearing the brunt of its impact.”
…
However, the Scottish Greens claimed Mr Sharma had “repeatedly voted against climate action” in the Commons and said his appointment threatened to “derail” progress on the issue.
In 2016 he voted against the energy industry being required to put in place a strategy for carbon capture and storage and also against setting a decarbonisation target for the UK.”
His degree is in Applied Physics with Electronics, so he might be biding his time.
He is soon to decide on Graveney’s 1000 acre solar power plant with associated 700MWh Li-ion battery (I can hardly believe those numbers but they are correct). Let’s hope Physics wins over his desire to keep his job by pleasing Boris’s squeeze and have something to claim credit for at COP26 but I doubt it.
Since he’s also an accountant he should take a look at the ridiculous cost of all the UK’s empty climate posing.
There is also the threat of cyber terrorism by making the 700MWh battery run hot and explode poisoning us poor locals or fluctuate enough to destabilise the grid. I fear that we shall be a sacrificial lamb to buy off critics at COP26. savegraveneymarshes.org has the whole gruesome tale.
Solar and wind can’t compete with gas that’s this cheap.
https://www.bnnbloomberg.ca/solar-and-wind-power-can-t-compete-with-gas-that-s-this-cheap-1.1390308
Solar and wind cannot compete, period.
That’s in North America. They have a glut of gas because they don’t have as much export capacity as gas they’d like to export. The USA didn’t have any export capacity before 2016.
The gas glut is global thanks to the LNG market. 20.69 p/therm on the UKP NBP spot or about $2.70/MMBtu is not much higher than the US price (and way below post Fukushima prices which soared well over $10/MMBtu in Europe), and allows CCGT to make money (before green taxes) at less than 2p/kWh or £20/MWh. Coal is also quite cheap at $52/tonne CIF Rotterdam for API2 quality. A more efficient coal station could actually be cheaper than CCGT before green taxes, but older less efficient stations work out to be slightly more pricey than gas.
Of course with the high winds across the continent we have been seeing negative prices. See here for instance:
https://www.energy-charts.de/price.htm?year=2020&auction=15m&month=2
Wind power IS predictable: it will fail.
I did some statistical analysis on this a few years ago (half hourly data between 29 November 2014 and 22 April 2016). Several things emerged:
High forecast wind generation often appeared to be curtailed by the time we see the actual generation. That is of course still very much (perhaps even more) the case, though how much of the curtailment is incorporated in the forecasts where it is due to transmission constraints such as the loss of the Western Link HVDC line can be hard to discern. Curtailment payments are growing as capacity increases, which is exactly what you would expect.
That aside, forecasting accuracy was not great. The average difference between initial forecast and actual was an over-forecast of 533MW (about 10% as a percentage error). The standard deviation of the forecast error was 935MW (about 36% on a percentage error basis) . The average actual outturn was 2,775MW (capacity monitored was about 8GW).
Weather forecasts good – ha ha, been doing a lot of landscaping work recently, weather has been critical, has been totally wrong 6-12 hours out numerous times.
I first started looking at forecast versus outturn of wind power over 10 years ago when I was giving evidence at several wind farm public inquiries. It doesn’t appear that the situation has changed much over 10 years. With the electricity produced being proportional to the wind speed cubed, it has always been obvious that there will be massive differences between forecast and outturn.
The biggest absolute differences I found were mainly due to curtailment, which seems to be a feature of high levels of wind output in the part of the turbine power curve that is essentially invariant to wind speed, because it’s maximum generator output.
Thank you for the h/t, sir. Fame at last! 🙂
On Nat Grid ESO site, you can find a forecast that explicitly says it is the day-ahead forecast.
For operational purposes, I think they continue to make forecasts much closer ahead than that, but whether they publish them is another matter. Forecast at 4 hours out would be an important window because there’s a lot of back-up available only at 4 hours notice.
There’s also a difference between an “average” forecast and a “99% certain it will be at least” forecast – something like the latter would be relevant to operational decisions to ensure back-up is available given the security standards they operate to.
They would need to take into account the reliability of plant as well as weather. You can see how the red line above looks like plant tripped out and back in again. You see that quite often.
When they double the amount of Wind Generation I can’t see the owners of CCGTs keeping the plants open and running without compensation, which will immediately make Electricity more expensive.
They will need to keep open just as much Gas fire generation as they have now, if not more than now.
One of the items on the agenda to get to zero carbon by 2050 is carbon capture. They are confident that it will be made successful before then but, as someone pointed out, if they are so confident why are they building all those renewables?
Of course it can be made to work, but you lose 25% – 30% of your generation thus making Electricity even more expensive.
When it does work you have to find some way of disposing of Tons of CO2, which is yet another expense.
Lunatics are running Asylum UK.
There are two other problems according to the CCC:
1) The CCS technology that we have cannot take all of the CO2 out
2) You also need to factor in the CO2 produced upstream in drilling, distribution etc of natural gas. This is most likely to be out of our hands in Qatar, Russia etc
Perhaps the answer is to ask Qatar to convert their lpg to hydrogen for us and dissolve the co2 in water, they are on the coast.
Each Q-max would carry just a third of the energy it does today as LNG. That’s the unavoidable physics of the low energy density of hydrogen. Much better to ship the methane first.
Perhaps it’s worth pointing out that after the August blackout it transpired that the Grid didn’t have a clue about how much embedded generation was contributing either before it started tripping out or afterwards – which made handling the other losses so much more difficult. So even when they don’t have the added complexity of a forecast they are completely at sea.
The E3C final report said “In total across the event, the loss of embedded generation is estimated to be in the range of 1300MW to 1500MW.” They don’t have a more precise figure, and even the one they do have was back calculated from data on frequency drops. To put that in perspective, the loss at Hornsea was 737MW and from Little Barford 640MW – so it was of the same order of magnitude as the two major losses or bigger, yet never received publicity.
A forecast is better than nothing I suppose, but how do they factor in the power output versus wind speed when this is a cube law. A forecast is just that but a variation would be expected and normal so the power output due to that variation is significant. My mathematics is not great but double wind speed and power goes up by eight times (And vice versa of course). Wind also is not constant much of the time and can be extrememly variable. This puts a strain on the CCGT\coal generators to balance the power levels. This can only get worse with time if the stated planned increase in wind generation is to be carried out.
It just seems to be heading for a real crisis in due course? Are there any power engineers on here who can expand on my thoughts?
The cube law is just the theoretical content of the wind. Practicality is that turbines produce nothing until the wind gets up to 3-3.5m/sec, and remain relatively inefficient until the wind gets up to 6-7m/sec. They then more or less follow the cube law at about 75% of theoretical maximum efficiency until wind speed approaches the level at which their generators are at maximum output, which will be somewhere in the region of 10.5-13m/sec. Designs vary according to expected distribution of wind speeds. After that, output is constant despite increases in wind speed, because it is limited by the generator: blades are progressively feathered to make sure they don’t drive it too hard. At higher wind speeds, usually above 20 or 25m/sec, the turbine cuts out, leaving the blades pin wheeling and not driving the generator. At 20m/sec, the efficiency can be around an eighth of that at 10m/sec. The cube law doesn’t apply at all in the region of maximum output.
Hello Idau,
so for a lot if not most of the time the machine is generating it will obey the cube law and my thought is true. Even in feathering mode it probably fluctuates in power as the mechanism will cycle unless the wind is very steady in speed.
Well, no, not really. Models of wind speed distribution usually use the Weibull distribution: there’s a really excellent explanation of all the theory and calculations that takes a bit of time to work through, including some sample turbine characteristic power curves starting here:
http://drømstørre.dk/wp-content/wind/miller/windpower%20web/en/tour/wres/weibull.htm
Probably the best way to look at reality is to look at actual output distributions. Here’s 7 days of action at part of the Hornsdale wind farm in South Australia (the one with the big battery! – although this is just the wind farm):
http://nemlog.com.au/show/eunitweather/7d/?k1=HDWF1&k2=PORT%20PIRIE%20AERODROME%20AWS
If you do a scatter plot of the output against the wind speed you will see that there is no obvious correlation at all, due to the extensive curtailment imposed – sometimes partial, and sometimes total. It has to be said that even adjacent wind speed measurements are not often very useful, even when there is no curtailment going on.
Perhaps it makes more sense to look at the distribution of total wind farm output across the country. I found that was roughly trapezoidal in shape: low outputs are slightly more common than higher ones, but higher outputs tend to be curtailed. This is evident when you look at the forecast data compared with actual.
Well, there is ample evidence that weather/climate is unpredictable by nature: that is why wind, ambient temperature, storms, clouds et al are unpredictable, all of them varying at random and sort of all over the place.
To my mind, the idea of generating solar and wind power is a means to reduce the usage of fossil fuels and thus reduce as much as possible the toxic pollutants from such power generating plants: it may be a bit more expensive, but well worth because of the reduction of deleterious health effects of such pollutants….