Wind power variability

In his response to the article on wind power written by Mark Lynas and me, Professor Gordon Hughes says that gas turbines need to be kept running because the amount of electricity generated by wind varies so rapidly. This short note examines the actual variability of wind power generation over the last three months and compares it to the variability of total demand for electricity. I show that the demand for power is typically over ten times as variable as the supply of wind generated electricity. The point is this: if the National Grid can cope with large half hourly swings in the demand for electricity, then it can cope with the erratic supply from wind farms. Because supply and demand must balance on an electricity grid, swings in demand have exactly the same impact as similarly sized variations in supply.

I analysed the electricity produced each half hour from 2nd July to today, 2nd October.


Degree of variation between one half hour and the next Number of instances 2nd July to 2nd October
Less than 50 MW 2698
51-100 MW 1087
101-200 MW 550
201-300 MW 91
301-400 MW 11
401-500 MW 2
501-600 MW 1
601-700 MW 1
Average variation 52 MW


The average variation in wind output was 52 MW. I then compared this figure to swings in total electricity demand in the same period. The average variation in demand was 678 MW, more than ten times as great as the average variability of wind output. In fact, the maximum variation in wind output between adjacent half hours (674 MW) was less than the average variation in total demand. The maximum half hourly swing in demand was almost four gigawatts, or about six times the maximum variation in wind power output.

The National Grid can cope with much more rapid changes in the supply or demand for electricity than are currently ever likely from the use of the current number of wind farms.

Some criticisms can be made of my simple comparison. First, I am using data from a time of year when wind power generation is relatively low. In winter, variability of wind generation will be greater. But variations in demand will also be much greater in the dark months of December and January. Second, it can be contended that demand variations are more predictable than swings in wind power. This point has some validity: demand moves up and down each day according to a relatively predictable pattern. However unforecast variations from the predicted level of demand can and do occur and these will be far greater than today’s wind variability. Weather forecasting allows good prediction of when power will increase or drop. Third, what is true today may not be true when the UK grid has to cope with perhaps five times as much wind power as at present. However even if we multiply the maximum half hourly variability of wind power in the last three months five-fold, we would still see less variation in supply than the maximum variation in demand experienced over the last three months.

Wind does not impose on the National Grid a substantial extra burden to balance supply and demand than exists already.


  1. Oliver Tickell’s avatar

    Good work chaps. Nothing like data to put some silly argument to rest. Odd thing here – your critic is a professor! Professor of what I wonder – medieval Persian literature perhaps? 18th century needlework? He obviously knows sod all about energy.

  2. Francis Macnaughton’s avatar

    Prof Hughes is an economic expert and seems to be linked to Nigel Lawson’s Global Warming Policy foundation.

    I find the near real time Grid data available at the NETA site informative

    Especially seeing how the forecast wind compares with the actual and also the overall forecast Grid demand versus actual

  3. David Ward’s avatar

    A reasonably good presentation of the data. However to be above criticism, I think it should be made clear that the data you are using (I think) are half-hour averages of MW. I would be interested to see the short term variability – to compare that of wind with that of demand.

    Also I think your statement “However unforecast variations from the predicted level of demand can and do occur and these will be far greater than today’s wind variability.” They do indeed occur, but it is not clear to me that you have demonstrated the “far greater”; to be squeaky clean you need to do so. You could look at the variation of “demand forecast error”, which I think would be much closer to the variation of wind output.


    Dave W

  4. Rob Glover’s avatar

    A good and thorough analysis. One criticism that I can foresee however is the variation values are variations on the current, installed metered wind base of 5GW. As the base of installed wind power expands towards the 28GW 2020 target these absolute values of swing will increase in turn – a simple pro rata would suggest 250MW as an average, likely still far below demand swings, but it will no longer be a factor of 10 below the demand side swings.

  5. Scott Luft’s avatar

    I’d challenge you to think this through conceptually treating the cost of electricity as relevant – meaning efforts would be made to control costs if real inflation in the sector was significant.
    If the wind turbines are not expected to be productive at peak demand periods, that means there will be more generator capacity, and therein lies the issue. Who will develop next generation coal and gas plants to run at 10% capacity factors instead of 80% capacity factors?
    I think it is a highly questionable premise that a grid with any wind will, not could, but will, end up with as few emissions as a grid that took the base demand level, met that base load with what hydro is available, and nuclear, and developed the remaining need as efficiently as possible allowing greater capital investments in fossil plants (hopefully moving to carbon capture and combined heat and power).
    The statements about demand variability being of greater magnitude than the variability in the wind supply is not well thought out. The two do combine at times; ‘must take’ renewables are essentially negative load. Sometimes they lessen load a lot, sometimes a little, and sometimes they will be moving from a lot to a little as actual load is also increasing.
    The argument has been that the need for companion dispatchable generation makes nuclear power and renewables incompatible, but I suggest it’s true of any high capital cost generation.

  6. Paul Dodgshun’s avatar
    Date: 28/09/12
    Professor Gordon Hughes responds: 
    5. This is the first point where the Goodall-Lynas evidence is incomplete. It relies upon data about the plants which are supplying electricity to the grid. It takes no account of the CO2 emissions of plants that are operating as spinning reserve. For simplicity, let us suppose that all spinning reserve is provided by gas combined cycle plants (CCGTs). If changes in wind output are balanced by changes in the level of spinning reserve, then the total amount of gas that is burned – and, thus, CO2 emissions – is completely independent of change in wind output. In terms of the Goodall-Lynas evidence, higher levels of wind generation displace gas generation one-for-one. But, there is absolutely no saving in CO2 emissions because the gas plants carry on running as before but they are just feeding less electricity into the grid. The reason for the error is that their figures take no account of what is happening in the parts of the electricity system that they have ignored.

    ‘If changes in wind output are balanced by changes in the level of spinning reserve, then the total amount of gas that is burned … is completely independent of change in wind output.’
    NO: This statement is flawed. ‘Spinning spare’ does not consume fuel. It is a lost opportunity to consume fuel and sell electricity, whilst being essential to the system. Use the ‘spinning spare’ and the CO2 emissions rise as the fuel burn rate rises.

    ‘… , higher levels of wind generation displace gas generation one-for-one.’
    YES: For grid connected units this statement is true.

    ‘But, there is absolutely no saving in CO2 emissions because the gas plants carry on running as before but they are just feeding less electricity into the grid.’
    NO – higher levels of wind that increase wind generation reduce gas plant power in this scenario, so CO2 emissions are saved .

    The electrical power generated in a fossil fired generator is proportional to the rate at which fuel is burned. Double the fuel flow and the electricity output doubles. ‘Spinning spare’ is deliberately created by not taking a unit to full power. To raise power further, and thus reduce ‘spinning spare’, requires the fuel flow rate to increase.

    This is no different to making a car go faster by pushing down on the accelerator. Push the accelerator, the fuel flow rate to the engine increases; that increases the engine power which makes the car accelerate to a higher speed.

    I am not sure what this does to the rest of the professor’s argument; I stopped at this point.


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

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>