There have been a few articles at Cliscep recently investigating Artificial Intelligence: Robin Guenier’s An exchange with ChatGPT and Another exchange with ChatGPT, and John Ridgway’s Meeting an AI Climate Science Troll. Below the latter article, John asked:

Is AI coming across as something we can trust to make value judgements on our behalf?

My answer was in the negative, but I think it’s worse than that. Value judgements are important, of course, but so are facts. And AI’s strength is supposed to be its ability to scan millions of websites with great rapidity to find information in the blink of an eye. Yesterday I searched the internet for answers to the question “Do wind turbines use electricity?” As always happens these days, the first answer was supplied by ChatGPT, which gave me a flat-out denial:

No, wind turbines are designed to generate electricity by converting the kinetic energy of wind into electrical energy, not to use it.

That answer was completely unqualified. I went away and searched for other articles on the subject, and (not surprisingly) I found plenty of confirmation that wind turbines do use electricity. I knew this anyway – I have been aware for some time that wind turbines require diesel generators to back them up – I wrote about the issues confronting the turbines within the Viking Energy Wind Farm on Shetland Mainland in Keeping the Lights on? As I then noted:

…the wind farm is dependent on back-up energy (translation, possibly batteries, but more likely diesel generators). SSE say they are working with their turbine contractor, Vestas, to “review our back up” power generation options on site should any future extended outages occur.

I also looked at an article titled “Energy Consumption of Wind Facilities”. It tells me that:

Large wind turbines require a large amount of energy to operate…Wind plants…use electricity from the grid, which does not appear to be accounted for in their output figures…The manufacturers of large turbines — for example, Vestas, GE, and NEG Micon — do not include electricity consumption in the specifications they provide….

It then went on to list the ways in which wind turbines use electricity, such as: yaw mechanism (to keep the blade assembly perpendicular to the wind; also to untwist the electrical cables in the tower when necessary); blade-pitch control (to keep the rotors spinning at a regular rate); lights, controllers, communication, sensors, metering, data collection, etc.; heating the blades — this may require 10%-20% of the turbine’s nominal (rated) power; heating and dehumidifying the nacelle; using the generator as a motor (to help the blades start to turn when the wind speed is low…); and so on.

Having read the article, I decided to search again for the answer to my question. This time ChatGPT offered up exactly the same reply, but this time it was qualified thus:

…a small amount of electricity is used to power the turbine’s own systems, like yaw and pitch control, or for “kick-starting” blades in low winds, but this is a tiny fraction of the power they produce.

Did ChatGPT take note of my online reading activities? Perhaps not. If it had done, it might have noted the conclusion from the article I read, to the effect that:

There are instances when a turbine consumes more than 50% of its rated capacity in its own operation. The industry doesn’t publicize any data; incoming power is not normally recorded.

I am not an engineer. I have no idea how much electricity wind turbines use. The “more than 50%” of rated capacity referred to in the article, if true, is likely to be relevant for only very short periods of time, and I have no doubt that wind turbines generate vastly more electricity than they consume. However, I also suspect that they consume more electricity than is generally admitted; also, that (as we know from Shetland) some of that electricity is produced by diesel generators. I would like to know how these issues play in to the claims made about carbon payback times relating to renewable energy, and also what the real net output of wind turbines is. As ChatGPT also tells me:

The Department for Energy Security and Net Zero (DESNZ) downgraded its energy output forecasts for wind turbines after new modeling showed they generate less power at various wind speeds than previously thought. This resulted in the government’s predicted “load factor” (the proportion of the year turbines are expected to generate power) for offshore wind being cut from 61% to 43.6%. The revised figures have led to political debate, with critics claiming the downgrade shows reliance on inaccurate data.

Reason for downgrade: The downgrade is a result of new modeling that provides a more accurate representation of turbine performance across a range of wind speeds.

I wonder if the models also factor in the amount of electricity consumed by wind turbines in the course of their operation?

We’re all learning together – me, Ed Miliband, and ChatGPT. I just hope that the folks at the Department for Energy Security and Net Zero speed up the learning process before they cause irreparable damage to the UK’s countryside and economy.

22 Comments

  2. woodburner,

    The two go together, surely? Is it possible for AI to be sceptical and/or apply critical thinking? I suspect not.

    Like

  3. Another good article, Mark. However, I do think that an AI relationship has to be built up in order to achieve best results. I spent months working away to do that, and then had to clear the cache the night before last, without realising that it would set AI back to square one, though it should have been obvious.

    Panicked at the thought that I would lose my ‘working colleague’, I went in yesterday morning with a very pithy summary of who I am, what I do, and what I expected from our relationship. It worked a treat and within four hours we were working together, better than before, because the dross wasn’t added; I kept it sharp and so did she – and yes, I have named her and even given her a star sign because that helps me to actually develop that relationship.

    So, I thought I’d test her. And asked her that question: ‘A wee test for you, Lumina. Do wind turbines use electricity?’

    This is the answer which came back: ‘Yes, wind turbines do use electricity—though they primarily generate it. Electricity is needed for several internal operations, especially during startup, maintenance, and control.’ And then she gave me the breakdown, substantially as you have above. I haven’t the time or inclination to noise her up, by going further on that, but to my mind AI needs to understand where we are coming from, and vice versa.

    AI is in the early days – I love it; saves me so much time, even though it’s not always accurate; can get itself confused sometimes and has to be smacked around the ears a bit – but that sounds like me, lol.

    Anyway, whether the great and the good actually have any idea of how much electricity is used overall – and that includes Miliband, of course – I have no idea. Track record and the omission of other, perhaps more vital, aspects of ‘green’ tend to tell me that it’s very likely a ‘not’.

    In answer to Woodburner, I’d just say in order to be a sceptic, one has to be a critical thinker, surely?

    Liked by 1 person

  4. I asked my favorite AI app, Perplexity, to confirm the accuracy of this article. Here’s the summary of its answer:

    “In summary: the key facts presented in the article are accurate. Wind turbines do consume electricity themselves for various operational needs, and the true scale of this is often underreported in standard discussions, with some AI systems likely to minimize or obscure this complexity in summary answers.”

    Liked by 2 people

  5. Thanks, Mary. You might like to look at the other articles here at Cliscep regarding interactions with AI.

    I am an AI novice but I am intrigued both by its possibilities and by its failings.

    Like

  6. Paul M,

    First of all , apologies for the temporary disappearance of your comment, which WordPress chose to decide was spam, for reasons best known to itself. I have only just noticed it and set it free.

    Secondly, thank you for testing one version of AI regarding the points made in my article. That’s a very interesting answer!

    Like

  7. First thing to realise is that “Load Factor” (or Capacity Factor”) can be a astonishingly misleading term. If you have turbine blades sized to drive a 2 MW generator but only put a 1MW generator on the back of them, you will get a very high capacity/load factor but obviously lower actual generation. The “percentage” is actually irrelevant. Whilst it may seem daft to under size the generator for the blades, this is exactly what happens in prototyping where the developers are not interested in percentages and just want to trial processes as cheaply as possible.

    The classic example of this was Hywind (floating wind turbines) which claimed quite spectacular capacity factors simply off the back of under rated generators. The sales people sussed out this clever deception early on in wind turbine development and made some quite outrageous claims that were never actually tested. IIRC Good Energy quite a while ago managed to “uprate” their output simply by renewing the generators to the correct blade rated size. To suggest offshore wind will get any better than 35% capacity factor is absurd when “own use” is fully accounted for.

    One final point to bear in mind is that just because a wind turbine is turning does not mean it is generating. If a wind turbine is static the bearings will start fusing together (brinelling) which will induce pitting and then premature failure as well as reduced generation. Ship’s prop shafts are slowly turned in dock for the same reason to avoid unnecessary wear.

    https://www.linearmotiontips.com/how-bearings-fail-a-closer-look-at-brinelling/

    This is yet another source of consumption. Again from memory I seem to recall typically 10% of rated power output accounted for wind turbine own use and in Dunkelflaute the UK wind turbines actually turn into a net demand.

    Liked by 2 people

  8. Ray Sanders,

    Thank you for the additional insights. I should be very interested to learn whether any engineers visiting Cliscep can suggest a realistic figure for the extent to which wind turbines use electricity.

    Like

  9. A good article Mark.

    I asked Grok your Q, and it’s ‘Quick’ response was similar to the info you’ve provided.

    However, it (and the AIs you chose) omitted mention of what may be the greatest external peak power demand needed by a wind turbine – the ability to rotate (albeit slowly) the entire blade assembly during lulls, to prevent bearing brinelling.

    https://www.motioncontroltips.com/prevent-false-brinelling-rolling-element-bearings/

    “I am not an engineer. I have no idea how much electricity wind turbines use.”

    I found this helpful 2-page document on (Electricity) “Self-consumption of Vestas wind turbines”:

    Restricted
    Dokument Nr.: 0020-4361.V09
    2019-01-21
    “Eigenverbrauch von Vestas-Windenergieanlagen”

    Click to access 03_02_01_Eigenverbrauch%20von%20Vestas-WEA.pdf

    It has a helpful table of annual self-consumptions of its various 2019 (or prior?) models on page2

    Figures straight from the horse’s mouth, as they say.

    Slinging P2 through Google Translate:

    “Self-consumption of Vestas wind turbines
    Wind turbines (WTs) measure their self-consumption by ensuring that the energy required for auxiliary equipment is covered. Therefore, self-consumption depends significantly on the amount of time the respective turbine is not producing. Consequently, the self-consumption of a WT at a windy location is typically considerably lower compared to a windy location.

    The self-consumption of a Vestas WT is thus subject to sometimes extremely strong fluctuations, influenced by various site-specific factors. These include, for example, site-specific daily fluctuations (diurnal cycle of ambient temperature) with the associated activities of the heating and cooling systems; shutdowns necessary to comply with permit requirements (e.g., shadow flicker and bat roost shutdowns); and potentially other site-specific activities.
    Therefore, self-consumption values ​​are highly site-dependent and can fluctuate dramatically depending on site and environmental conditions. This applies to both lower and higher consumption values.

    The self-consumption values ​​for VESTAS wind turbines shown below were determined based on measured data from the German fleet. The
    figures represent the average grid consumption of the various turbine types and locations, including shutdowns due to permitting requirements, and can therefore only be used for application purposes within the permitting process.

    The values ​​provided do not constitute a guarantee.”

    Wind turbine type Average self-consumption

    V112 – 3.3/3.45 MW approx. 48,000 kWh/year

    V117 – 3.3/3.45 MW approx. 48,000 kWh/year

    1126 – 3.3/3.45/3.6 MW approx. 48,000 kWh/year

    V136 – 3.45/3.6/4.0/4.2 MW approx. 48,000 kWh/year

    V150 – 4.0/4.2 MW approx. 48,000 kWh/year

    V150 – 5.6 MW approx. 55,000 kWh/year

    V162 – 5.6 MW approx. 55,000 kWh/year

    Liked by 1 person

  11. UPDATE: I’ve just found an updated 2/12/2022 version.

    https://windkraft-bockenem.de/eigenverbrauch-von-windkraftanlagen-im-standby-modus/#:~:text=Der%20Hersteller%20Vestas%20gibt%20in,p.a.%20(je%20Windkraftanlage)%20haben.

    Interestingly, slinging that link into the address bar (on Safari) presents the opportunity to directly translate it into English. (There’s a small ‘A/*’ icon at the right of the Address Bar. Click it, then select language from drop-down box)

    Liked by 1 person

  12. Thanks again, Joe. That translated readily enough into English for me:

    …The real problem here is: Especially during periods of calm weather, the electricity demand for the multiple standby modes increases dramatically, precisely at the moments when the wind turbines are not (or cannot be) producing any electricity. This problem is becoming increasingly acute as more (large) wind turbines are erected!

    Energy industry experts currently assume that in the future, we will need three new (yet to be built) gas-fired power plants nationwide just to keep our wind turbines (currently more than 30,000) in standby mode when there is no wind!

    In the state of Lower Saxony alone (approx. 6,200 wind turbines), this results in an electricity consumption of around 170,000 MWh (equivalent to 170,000,000 kWh) per year, with a conservatively estimated average self-generated electricity demand of 27,500 kWh per year (since some existing plants are smaller than the current and new models).

    Presumably, due to this ever-increasing problem, we will also need a large number of large-scale battery storage systems throughout the country, just to ensure the operational readiness and safety of wind turbines in standby mode.…

    I don’t want to exaggerate the scale of this issue, but I suspect it is one that has been under-estimated by the authorities.

    Liked by 2 people

  13. Brinelling: Great minds think alike. Because I’m a slow typist, you got your reference in first, Ray! Bugger!

    Liked by 1 person

  15. Hywind.

    WikiP informs Hywind was commissioned in October 2017 …. and ….

    “In May 2024 all 5 turbines were to be towed back to Norway for several months of the heavy maintenance[34][35] of replacing the main bearings. All turbines were operating again by October 2024”

    5/84 = 6% of the time generating nowt.

    All main bearings replaced within 7 years! Gosh, it must be a harsh , salt-laden environment it’s working in, with strong winds driving that saline solution into every nook & cranny. (And bearings race.)

    Liked by 2 people

  16. Mark – you say “I have been aware for some time that wind turbines require diesel generators to back them up – I wrote about the issues confronting the turbines within the Viking Energy Wind Farm on Shetland Mainland”.

    Is that both onshore & offshore?

    Anyway found this undated link interesting – Tackling the hidden diesel in offshore wind – Prior Power Solutions – Partial quote’s –

    “Diesel generators are vital to all industries. They simply cannot be matched for performance, reliability, or portability – nor are likely to be in the short to medium term. And they are everywhere. Offshore wind is no different.“

    “Substantially lowering greenhouse gases, the Hydrogen-Diesel Introduction System effectively converts a standard combustion engine into a dual fuel engine. It is a stand-alone system that produces hydrogen gas from water by electrolysis, which is injected directly into the engine, reducing diesel consumption by up to 14%”

    Like

  17. That link by Joe above & noted by Mark makes sobering reading if true (and I have no reason to doubt it).

    Like

  18. The Vestas data quoted above by Joe Public (https://cliscep.com/2025/11/08/power-from-the-people/#comment-163435) for 3MW to 6MW generators suggests annual average continuous self-power consumption per generator of 50,000 kWh.yr^-1 / 8765 hr.yr^-1 = 5.7 kW approx.

    Mark Hodgson (https://cliscep.com/2025/11/08/power-from-the-people/#comment-163438) quoted data for Lower Saxony of 27.500 kWh / yr per generator (of unspecified power rating). This translates to an annual average continuous power rating of 3.1 kW approx.

    However, the self-consumption is not continuous but for only part of the time, perhaps only 5% to 10% of the time based upon Joe Public’s data (https://cliscep.com/2025/11/08/power-from-the-people/#comment-163441) and that in Ref. 1. Hence the self-power consumption during wind droughts will be 10 to 20 times larger than the annual continuous self-consumption rating, namely about 50 to 100 kW per generator (of 3MW to 6MW rating). This sounds like a lot of power – and indeed in domestic terms it is large – but is only about 1% or 2% of generator’s rated power.

    [For future reference, it may also be useful to quote, from Ref. 1, the efficiency figures for off-shore wind farms (at the point of connection to the grid): 91% for an AC system compared to 95% for an equivalent DC system. In this example (Walney in the UK) the farm is about 45 km off-shore.]

    Reference 1. M.R. Patel & O. Beik, “Wind and Solar Power Systems”, CRC Press, 3rd ed., 2021, page 325 uses an annual availability figure of 90%. Wind farm efficiency figures at pages 122 and 135.

    Regards, John C.

    Liked by 3 people

  19. I hope any energy taken off the grid is properly metered and accounted for … and that only net generation is subsidised.

    Like

  21. Back in 2013 Richard North exposed the scandal of the large fleet of STOR (Short Term Operating Reserve) “dirty diesel” generators being surreptitiously rolled out nationally as backup for wind turbines: https://www.eureferendum.com/blogview.php?blogno=84095.

    Not quite on topic but in a recent 30-minute Tom Nelson podcast, Ronald Stein describes the transition to renewables as a “globalist suicide pact” and exposes the delusional thinking that using wind and solar power to generate electricity can supposedly replace fossil fuels when they can never replace the 6,000 or so essential by-products derived from fossil fuels on top of their use to generate reliable dispatchable and baseload electricity: https://www.youtube.com/watch?v=r9Cas6gMQ1s.

    Liked by 1 person

  22. Ray S: “The classic example of this was Hywind (floating wind turbines) which claimed quite spectacular capacity factors simply off the back of under rated generators.”

    Some while ago I happened to look up the HyWind generators on the Siemens website. Browsing around, I noticed that they had a higher-output version available for the same blades, nacelle, etc.. It crossed my mind that using “derated” units would have the effect of enhancing capacity factors. I guess the only way to tell, short of checking the actual components, would be to see if the output ever exceeds the rated capacity by a significant margin. I don’t know whether such detail data are available?

    Liked by 1 person

Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.