This post gives estimates, for the electricity grid of Great Britain, of how much metered wind power would be produced in 2016/17 if there is a repeat of the coldest winter weather between 2007/08 and 2015/16. Of particular interest is the amount of wind power expected at the times of peak demand on the coldest (highest demand) days. It is found that the coldest days between 2007/08 and 2015/16 were associated with relatively light winds, such that wind power in 2016/17 would often contribute only around 1-2% of the highest peak demand. These results provide strong constraints on the minimum amount of dispatchable power required to meet peak demands in forthcoming winters.
The start date of the analysis is 2007/08 because that is the earliest date for which data is readily available for electricity demand and for wind power, both available from a National Grid website. The National Grid provide estimated data for embedded (non-metered) wind power (mainly from single turbines and small wind farms), and I have used this data to estimate the amount of metered wind power that would have been produced if the generators of 2016/17 had been present in past winters. See the NATIONAL GRID DATA page of this blog for the data download link and a discussion of the accuracy of this estimation technique.
Electricity demand data itself is a good way to find days of high demand, but it is heavily modulated (reduced) during weekends (including Fridays to a certain extent) and the Christmas/New Year holiday period. Thus, I have used temperature data as the starting point for finding days of actual or potential high demand. The temperature used was the daily maximum of the Central England Temperature (HadCET) series (plotted in the following figure), with a threshold of zero degrees C:
The figure above identifies 4 winters with relatively cold (for GB) days: 2008/09, 2009/10, 2010/11 (very cold), and 2012/13. The estimated metered wind power at 2016/17 levels at peak demand time for these 4 winters are shown in the following figure, identifying the coldest days with blue squares:
Whilst there were some cold days with substantial wind power it can be seen that most of them fell on days with relatively light wind, some at deep wind lulls. The rest of this post provides a detailed analysis of demand, wind power and temperature for the lowest-wind cold days, but I give here the significance of these results. In the absence of proof that recent cold winters were exceptional, and are very unlikely to be seen again, wind power cannot be relied upon to contribute much to meeting the highest peak demands in future winters. See the “policy conclusions” section at the end of this post for further discussion.
Wind power may save the day during blizzards but, with insufficient dispatchable power, is unlikely to do so during the calm cold weather that has been seen in recent cold winters.
Analysis of cold low-wind days
The remainder of this post shows demand and weather data for the lowest-wind cold days.
The wind data for December 2008 to February 2009 is repeated in the following figure, together with the daily peak demand:
The cold day of 31st December 2008 fell during the Christmas holiday period, so its potential demand (if it had fallen on a working day) is unknown, but its low wind credentials are confirmed from the following synoptic chart for 31st December:
Finally, the following figure shows half-hourly weather data measured at Cambridge (source: see the WEATHER DATA LINKS page of this blog):
The 30th December was a sunny day, which probably allowed the temperature to drop substantially overnight, aided by the light winds inhibiting replacement and mixing of the resulting cold dense air, which then failed to get much warmth during the overcast day that followed.
The wind data for December 2009 to February 2010 is repeated in the following figure, together with the daily peak demand:
The following synoptic chart for 7th January 2010 confirms the light winds:
The following half-hourly weather data measured at Cambridge shows several days with temperatures failing to rise much above zero degrees C:
The wind data for November 2010 to January 2011 is repeated in the following figure, together with the daily peak demand:
The cold day of 28th November 2010 fell on a Sunday, so its potential demand is unknown, but its low wind credentials are confirmed from the following synoptic chart:
The half-hourly weather data measured at Cambridge confirms the low temperatures during 28th November 2010, despite many hours of sunshine:
The cold days of 6/7th December 2010 both fell on working days and resulted in high demands, with the following synoptic chart confirming light winds:
The half-hourly weather data measured at Cambridge confirms the very low temperatures on 6/7th December 2010:
The cold day of 20th December 2010 fell on a working day and resulted in high demands, with the following synoptic chart confirming light winds:
The half-hourly weather data measured at Cambridge confirms the very low temperatures on 20th December 2010:
The wind data for December 2012 to February 2013 is repeated in the following figure, together with the daily peak demand:
The cold day of 12th December 2012 fell on a working day and resulted in high demands, with the following synoptic chart confirming light winds:
The half-hourly weather data measured at Cambridge confirms the low temperatures on 12th December 2010:
The cold day of 16th January 2013 fell on a working day and resulted in high demands, with the following synoptic chart confirming light winds:
The half-hourly weather data measured at Cambridge confirms the very low temperatures on 16th January 2013:
Capacity credit for wind power has developed into an academic field of study, but the results presented here suggest to the author that simple engineering rules-of-thumb are adequate.
There have been several examples in recent years of cold calm weather producing very high peak demand for electricity, with wind power at 2016/17 levels contributing only a very small percentage of that peak demand. I am reluctant to quantify the contribution of wind power to the highest peak demands, as that might lend it an undue level of credibility, not justified by the very short time period examined. There is no obvious reason why even lower wind power contributions might not occur.
Thus, wind power (like solar PV) should not be considered when determining how much dispatchable power is needed to meet peak demand. The small percentage contribution of wind power to the highest peak demands also suggests its neglect for economic reasons. The cost saving to consumers of providing (say) 99% rather than 100% of the required dispactable power is very small, less than 1% because there are many other costs besides that of providing the generating capacity.