Skip to content

Gridwatch’s Leo Smith On Renewable’s Instability

April 30, 2020

By Paul Homewood

  h/t Douglas Brodie


Leo Smith, who runs the Gridwatch website left this comment on WUWT the other day, following up on John Constable’s article here.

Being an engineer, Leo knows his stuff inside out, and his comment is worth reposting in full:


This is a good article but from the comments it seems that some of the important detail has not been fleshed out enough.
First of all let me say something about solar power in the UK.There is quite a lot of it, but none of it is metered centrally by the grid, and, as far as I know none of it is under central control. If you look at the demand curve graphs you will see that there is a big midday notch. I have been monitoring, via my website, Gridwatch, ( the same data as is used here, since 2011. There never was a midday notch until solar power came along, and this ‘notch’ is actually the effect of millions of domestic solar panels and a few commercial solar farms kicking in and feeding local demand. As far as the central monitoring is concerned this represents a fall in ‘seen’ demand and so ‘the notch’. A couple of years back Sheffield University, in a zeal for all things renewable, starting publishing their estimate of total GB solar power, based I think on sampling a few sites who agreed to share their output figures. I now incorporate that data with suitable caveats on the same website and database.

This solar power is absolutely ‘wild’ on the grid. As far as domestic panels go, there is absolutely no way to control the output centrally at all. I am less sure about the few commercial solar farms as to what their contractual arrangements are and whether (like wind) they can be ‘constrained off the grid’.

Wind farms are similarly wilful in their output, although both solar and wind are reasonably predictable to within a couple of GW several hours ahead, and a couple of GW is something the grid should be able to cope with – typically, if there is a massive and sudden shortfall if a power station trips out, the pumped hydro at Dinorwig and the small hydro sets get an alarm and can be up to power in seconds. So long as they can ‘hold the fort’ for half an hour or so without running out of water, that is enough to get the gas sets up and running. Similarly a sudden drop in demand is catered for by the inverse – hydro stops and then gas sets are shut down.

Contrary to popular belief not much ‘spinning reserve’ is kept online. I suspect that what actually happens is that the hydro plants are used to modulate very small and rapid fluctuations in demand over the timescale of minutes.

This however is not the instability which is the matter of concern here, what is a matter of concern is the grid frequency – nominally 50Hz± not a lot. Traditionally this was held to 50Hz exactly on average over the day so that synchronous electric clocks kept proper time, but minor fluctuation was tolerated. And a major part of that stability was inherent in the design of the generating plant – massive turbines and rotors in the alternators held to 3000rpm (for 3 pole designs) represent a surprising amount of stored kinetic energy. More than the average battery installation for sure, and in the case of – say – lightning striking a main grid link and shorting it – this is what supplies the overload current until the arc self extinguishes. Then of course we have electric trains which are constantly stopping and starting with power flows to and from the grid (most these days push braking energy back into the grid) .

This is all well and good until – because renewables have priority on the grid – there is hardly any conventional generation on the grid at all. Peak solar and a bit of wind is capable of putting over 16GW onto the grid and typically when Europe has a surplus in summer, we will be importing another 4GW via the undersea high voltage DC links. So that’s 20GW of ‘unconventional’ generation on a grid whose maximum total demand is in the middle of the day scarcely 30GW. Typically there will be a 3GW of wood burners at Drax plus about 6-8GW on nuclear power on the grid. And all the coal, gas and hydro shut down.

And this is where the problems start to get nasty. All those renewables, and the DC power under the sea, feeds the grid via inverters. Electronic circuits that chop the DC and shove it through high frequency transformers to generate an approximate 3 phase sinusoidal waveform that is fed to the grid via chokes and capacitors to improve the waveform . And this has to be in phase with the grid and locked to its frequency.
How do they know what the frequency is? They monitor it of course.

And what happens if the frequency, now hardly controlled by spinning turbines, drops due to a temporary overload?


And instead of 2GW lost from the grid, because a reactor has tripped, that’s 20GW, and a total cascading blackout across the entire country.
That is the problem that is concerning engineers, myself included.

You may recall that during last summer’s blackout, arguably the biggest problem was how embedded renewable generation suddenly dropped off, as the system frequency fell.

It would appear from John Constable’s analysis of the official investigations that this sudden loss of generation was never fully explained.

What we do know though is that problems of this nature will become progressively worse as more and more renewable energy comes on board.

  1. Joe Public permalink
    April 30, 2020 11:51 am

    A superb analysis, which ought to be required reading by all renewables-pushers.

  2. Patsy Lacey permalink
    April 30, 2020 11:53 am

    Compare this analysis with Ambrose Evans-Pritchard’s green wash in today’s Daily Telegraph

  3. jack broughton permalink
    April 30, 2020 12:17 pm

    I wonder if the Grid have calculated the allowable fraction of unreliables / asynchronous-generation based power that can be tolerated by the grid. We used to think 40% unreliables was the maximum, but I don’t know if that was an opinion or a proven value.

    Another factor in unreliables, that is conveniently overlooked, is the cost of back-up and providing power during the unreliable troughs. The CEGB used to penalise suppliers who committed to provide say 1 GW, then could not when asked. We now reward them and even pay them “constraint payments” when we can’t take their power: total madness.

    • Gerry, England permalink
      April 30, 2020 2:07 pm

      It was at 50% when it went pop last year. Loved to have been a fly on the wall to witness how they took the possible collapse of the whole grid.

  4. Bloke no longer down the pub permalink
    April 30, 2020 12:17 pm

    I’m sure someone will correct me but I thought that at grid scale, power was converted from DC to AC via a mechanical link, ie a DC motor turning an AC generator. If that is the case, for example where power comes into the UK from Europe, why can’t the speed of rotation be controlled as easily as from a thermal power station?

  5. Dick Goodwin permalink
    April 30, 2020 12:29 pm

    What a fantastic article, can we send it to the Gaurdian? I wish I had a quarter of Leo’s knowledge on this subject.

  6. April 30, 2020 12:44 pm

    thank you very much for re-posting this comment which i missed on wuwt for some reason. there is a lot of good information in there. I will save this link of course. It’s gold.

    again, thank you.

  7. Geoff B permalink
    April 30, 2020 1:54 pm

    I graduated 50 years ago with a degree in electrical engineering, Power generation and grid systems was about 15% of the syllabus. (a lot of Physics and Materials as background) a lot about instrumentation (measuring things) ..Computers and semiconductors were the in subjects, so I went with the money and creativity …Computers, in fact Disk drives and Magnetic tape storage systems, banks of whizzing reels of tape, like you see in old James Bond Films. I then went into tape recorders and hi fi, finally ending up making fuses.

    Nowadays I wonder how many electrical engineers actually understand the complexity of the AC grid system that Merz instigated. The concept of “power factor” and “reactive energy” (VARS) are not really well understand, as well as the requirement for spinning mass kinetic energy for frequency stabilisation. Now that semiconductors are used to convert the DC from offshore wind farms and the interconnectors, these devices have a very low I2t (I squared t) that means very little thermal mass so in fault conditions they heat up and burn out very quickly, quicker than conventional fuses and circuit breakers can trip. So they are protected very closely by “intelligent trips” as soon as the frequency starts to change they trip in less than 1 cycle (20mS).

    The worry is a cascade trip which is described in the article.

    BUT the big problem is if a large section of the grid goes down with a cascade trip, lets say London and the south then because of the capacitance of the overhead grid lines and the fact that its a “cold load” restart needing the ability to withstand a mammoth overload for up to 10 minutes as everything restarts. You cannot start it with the semiconductor inverters, you have to use spinning mass kinetic energy. Basically Coal or Nuclear which boil water make steam and spin a mammoth hunk of steam turbine.. However nuclear is closely controlled for obvious reasons and does not take well to overloads, they generally trip out fairly early. So its Coal and while our green loonies are crowing about not using any for 18 days be aware with only 3GW left its going to take a fair time to restart the grid. Its really a lot worse than made out.

    • mikewaite permalink
      April 30, 2020 3:55 pm

      What a pity this article and Geoff’s contribution above will never be read by Ambrose Evans-Pritchard whose latest piece of nonsense has been dissected in a later post here.

    • April 30, 2020 9:18 pm

      I understand this topic and am in touch with JC, but find your explanation perhaps the clearest in plain English I have seen. This all happens before even Dinorwig can start filling the gap, one cycle sensitivity to frequency perturbations is far too senitive to transient events to be a significant part of mainstream grid supply. The inertia must be there to provide grid stability.

  8. April 30, 2020 3:06 pm

    I made a copy of the comment when I fest read it on WUWT. I am already making use of it in an article I am writing.

  9. John Cullen permalink
    April 30, 2020 9:40 pm

    There is a minor typo in Leo’s text. The 3,000 rpm machines are in fact 2-pole machines (and not as written 3-pole machines) since f = np where f is the frequency in hertz (Hz), n is the rotational speed in revolutions per second (i.e. here n = 3000 / 60 = 50 rev/sec), and p is the number of pole-pairs (i.e. p = 2 poles/2 = 1). Thus f = 50 x 1 = 50 Hz in this case.

    By contrast, I understand that the French run many of their larger generators at 1,500 rpm because they are 4-pole machines and so for them f = (1500/60) x (4/2) = 50 Hz . However, because their grid is independent of that in the UK the two grids are connected by high power HV DC links which thereby does away with the issue of synchronising the phase angles of the two 50 Hz grids but introduces, via the associated thyristors and smoothing circuits, all the asynchronous problems that Leo wrote about.


  10. Iain Reid permalink
    May 1, 2020 10:42 am

    This is from The Guardian (Australian, I think?).

    “Australia already has the technical capacity to safely run a power grid in which 75% of the electricity comes from wind and solar and, if it gets regulations right, should occasionally reach this level within five years”

    The article, written by the Australian Energy Market Operator, says that the only limiting factor is electricty markets and grid congestion.
    It will be interesting to see how they get on?

    I did visit Jo Nova’s site to see if she had a view but it is all corona virus there at the moment.

    • Dave Ward permalink
      May 1, 2020 2:29 pm

      “Australia already has the technical capacity to safely run a power grid in which 75% of the electricity comes from wind and solar”

      This post suggests otherwise:

    • Stuart Brown permalink
      May 1, 2020 3:44 pm

      Typical lack-of-comprehension breathless reporting, I reckon. This is what the report actually says, (where Zone C is in a graph of demand against % renewable generation & only gets to 75% if the grid is at max demand):

      “If recommended actions are not taken to address the regional and NEM-wide technical challenges identified in this study, the identified operational limits will bind and constrain the output of wind and solar resources. This would limit their maximum contribution at any time in the NEM to between 50% and 60% of total generation.

      If recommended actions are taken, the NEM could potentially be operated securely out to the beginning of Zone C by 2025, with up to 75% of total generation coming from wind and solar resources at any time.

      Operation in Zone C, with up to as high as 100% of wind and solar generation operating securely at times, is theoretically achievable in future. This would, however, require more advanced methods of system operation, coupled with provision of essential system services to ensure adequate system flexibility, frequency, and voltage management.”

      Nowhere does it say such a thing can be achieved today, but I would agree it’s barmy to suggest it’s even thinkable! I’m surprised the press report didn’t just pick the 100% figure.

      • It doesn't add up... permalink
        May 2, 2020 6:54 pm

        As soon as you start to reach about 60% of energy supply over time being from renewables you start to run into the need for extensive curtailment or massive storage, in addition to the short term problems of grid stability when operating at high levels of renewables output. Both escalate the costs quite fast, and yet you still need to be providing 90+% backup in dispatchable form for when the sun doesn’t shine and the wind doesn’t blow.

      • May 2, 2020 7:13 pm

        Beat me to it!!

        See my new post!

    • Stuart Brown permalink
      May 1, 2020 3:47 pm

      … what the AEMO study actually says… as opposed to the press report.

  11. DaleC permalink
    May 2, 2020 1:28 am

    I have prepared a detailed set of charts of the GridWatch data, currently up to 12Apr20.
    For just the charts, see

    Click to access GridwatchUK_Demand_and_Supply.pdf

    For my discussion thereof, see

  12. JoeSoap permalink
    May 3, 2020 1:25 pm

    And what happens if the frequency, now hardly controlled by spinning turbines, drops due to a temporary overload?

  13. SoapySoutar permalink
    May 3, 2020 1:43 pm

    And what happens if the frequency, now hardly controlled by spinning turbines, drops due to a temporary overload?
    You mean a temp rise in demand to cause system frequency to drop ?

    What about system frequency rise due to a temp. rise in renewable generation (or loss of demand) ?

Comments are closed.

%d bloggers like this: