I was contacted asking had I made a basic error? Yes I had. But it is still true that wind and solar PV *capacities* are reported on the basis of fantasy load factors defined by legislation in 2010
I have been doing some analysis on the DUKES numbers. Here are some observations.
1) Taking 2021, the capacity factors implied by table 2 (using the end of year capacity not the average) and the stated generation are: Onshore wind - 22.9%; Offshore wind - 36.0% (Combined wind - 28.9%); Solar - 9.9%; Hydro 33.2%. These are pretty close to expected so I think we can presume that the generation values are accurate, although footnote 8 signals that estimates based on the estimated (inaccurate) capacity factors may have been used if actual data were not available.
2) Again using 2021 values, if we apply the stated de-rating factors (Hydro 0.365; Wind 0.43, Solar 0.17) to the stated Installed capacity values, we get a total of 22,234 MW, which is close to the Declared Net Capacity of 22,393 MW.
3) If we substitute the calculated capacity factors in 1) above for Combined wind, Solar and Hydro instead of the 2010 estimates, the total Declared Net Capacity would be 17,610 MW, which is 20% less than the stated DNC. So, the DUKES DNC values overestimate generation from those sources by 25%! If the DNC values are used for planning, that is certainly enough to generate some false optimism.
4) I agree that the average Installed capacity for any year is a better value than the end of year but without knowing when any particular capacity actually came on line, we don't know how much better.
I'd still like to understand why there is such a large gap between the DUKES generation numbers and Gridwatch. It would be useful to know how both sets of numbers are compiled.
Beware that Gridwatch only covers wind visible to National Grid. It excludes a lot of embedded wind connected at distribution levels. BM Reports does have a report that covers estimates of embedded wind, but the accuracy of that is questionable. It seems strange that your domestic smart meter could be monitored live every half hour, but they fail to do that for generation. Eventually meter readings are fed into the reconciliation of who gets paid what, but public visibility of the detail is missing.
One way to get some handle on the actuals, even if only at monthly resolution and somewhat in arrears, is to look at OFGEM's database on REGO certificates, which is the basis for much of the REF data. That is also a way of getting a better estimate of the timings of new capacity additions, at least for major projects. They do tend to ramp gradually as more turbines are added.
Of course there are questions as to how you should treat things like the Rampion wind farm being offline for many months because of cable failure. The same applies to interconnectors, with the fire at Sellindge on IFA1 being one example. The loss of Western Link HVDC for protracted periods on several occasions also will have resulted in much more wind curtailment in Scotland. In reality, we probably need to add back curtailed generation to understand the true levels of surpluses on windy days. The implication would be even more curtailment and/or storable surplus.
All good points. Can curtailed generation be estimated from the payments made or is it not linear? Are the data available?
It is of course one of the huge problems with wind that there will sometimes be too much so we will all have to pay for it to be curtailed. And the more capacity there is, the more that will happen, absent sufficient storage.
I suspect that curtailment / constraint payments are non-linear depending on the marginal market in that interval.
Wind proponents say that, because wind is free, it doesn't matter if we have more and more of it. However, all the while that consumers are on the hook for curtailment / constraint payments, this just adds to our bills. A knotty conundrum.
Yes, Gridwatch states that wind omits some generation, but it's otherwise is so convenient to use that data and the 'correction factors' derived from DUKES are so consistent that I'm happy to live with the approximations. I am, after all, seeking to understand the big picture not every teeny detail.
I've seen argument on LinkedIn forums by experienced power system people that wind and solar PV act as negative demands. Currently I'm persuaded by that argument.
It then begs the question, which if any of the parameters are 'accurate'? GB doesn't have a massive electricity meter for aggregate demand so that key parameter must be derived somehow. Again, not important for the big picture.
Currently curtailed wind is curtailed for good reasons, either local grid capacity is insufficient or overall surplus generation would unbalance the grid. Power storage able to provide meaningful extra demand would be immense as I've posted numerous times.
Although the arithmetic from the grid point of view adds up if they view embedded generation at the distribution level as negative demand, it can throw up surprises, as it did during the August 2019 blackout, when they discovered that knocking out some demand also knocked out embedded generators, and that the grid disturbances caused embedded generation to trip off line, increasing the requirement for backup generation substantially above what they could see in the control room: they had underestimated the risks. They could get caught the same way if they don't keep an eye on Continental grids and suddenly find several interconnectors at risk of tripping out. It is really much better to model the whole thing properly.
It is important to look at interconnectors too, now they are becoming such a significant slice of the supply/demand picture. I tried to put it all together in this chart:
Here you can see the pumped hydro and interconnector flows along with the various sources of generation. Underlying demand is derived from the net export/pumping flows being taken off the overall supply. It becomes clear that we ran extra gas generation to provide exports mainly to France over much of the year, to help them out with their nuclear shortfalls, and as a surrogate way of exporting gas because the gas export lines were full but we could land extra LNG. The role of gas as the first line responder to wind intermittency also becomes clear.
Hello 'It doesn't add up...' - your christening must have been interesting (joking).
As a Chemical Engineer not a Power Engineer I'm interested to use the data - Gridwatch is v convenient - to develop my understanding of the feasibility of 'decarbonisation'. How big the non-fossil sources of energy *are* versus demand, and how the big picture changes if the non-fossil sources are made multiples bigger.
The nitty-gritty of control systems, surge protection, trip robustness, etc. are beyond my area of expertise. [I read the Energy Emergencies Executive Committee (E3C): Final report into the GB power system disruption on 9 August 2019 and understood what it said at a superficial level but not deeply.]
Your chart has the same issue I've encountered with my analyses: it's complex. But yes, I saw that there was a lot of power going *to* France - hopefully GB got paid a good rate for it all.
Your comments of 02March arrived while I was on hols, I've only just realised I never responded.
I agree your points 1) & 2), and very much agree with your point 3) as it's the point I was making :)
Re: your point 4) I think I agree with the DUKES approach to estimating load factors; it should get much the same answer with an order or two magnitude less effort. As to the latter part of your comment, we know from the Gridwatch site (if you hover your cursor over that dial) that the displayed wind value is that which is metered, and that there's another ~30% unmetered. I assume that DUKES has methodology to estimate that extra. Them being the official statistician, I would hope so. The differences for solar PV between DUKES, Sheffield Uni and Gridwatch I cannot explain, but the factor is about constant across the available whole years.
To me, the most annoying fact is BEIS's narrative-based comparisons without providing Tables of Data that would enable its audience to easily compare the different variables and over time.
E.g. Year vs Year; Energy sources; Generation mixes etc, etc.
This is especially annoying because DUKES is 'UK', yet doesn't apportion data England / Scotland / Wales / Northern Ireland. (Or even Britain / Northern Ireland)
We in Britain have our grid & our data, and in real time. Northern Ireland has its SONI data. Naturally, they're not easily comparable.
I've been kicking myself for making such a basic error with the averages, and you helped cheer me up. But I'm right about that Note 6, it's horrifying if HM Gov policy is being guided using those unrealistic fixed load factors, especially for solar.
Hello Chris.
[Previous comments deleted]
I have been doing some analysis on the DUKES numbers. Here are some observations.
1) Taking 2021, the capacity factors implied by table 2 (using the end of year capacity not the average) and the stated generation are: Onshore wind - 22.9%; Offshore wind - 36.0% (Combined wind - 28.9%); Solar - 9.9%; Hydro 33.2%. These are pretty close to expected so I think we can presume that the generation values are accurate, although footnote 8 signals that estimates based on the estimated (inaccurate) capacity factors may have been used if actual data were not available.
2) Again using 2021 values, if we apply the stated de-rating factors (Hydro 0.365; Wind 0.43, Solar 0.17) to the stated Installed capacity values, we get a total of 22,234 MW, which is close to the Declared Net Capacity of 22,393 MW.
3) If we substitute the calculated capacity factors in 1) above for Combined wind, Solar and Hydro instead of the 2010 estimates, the total Declared Net Capacity would be 17,610 MW, which is 20% less than the stated DNC. So, the DUKES DNC values overestimate generation from those sources by 25%! If the DNC values are used for planning, that is certainly enough to generate some false optimism.
4) I agree that the average Installed capacity for any year is a better value than the end of year but without knowing when any particular capacity actually came on line, we don't know how much better.
I'd still like to understand why there is such a large gap between the DUKES generation numbers and Gridwatch. It would be useful to know how both sets of numbers are compiled.
Beware that Gridwatch only covers wind visible to National Grid. It excludes a lot of embedded wind connected at distribution levels. BM Reports does have a report that covers estimates of embedded wind, but the accuracy of that is questionable. It seems strange that your domestic smart meter could be monitored live every half hour, but they fail to do that for generation. Eventually meter readings are fed into the reconciliation of who gets paid what, but public visibility of the detail is missing.
One way to get some handle on the actuals, even if only at monthly resolution and somewhat in arrears, is to look at OFGEM's database on REGO certificates, which is the basis for much of the REF data. That is also a way of getting a better estimate of the timings of new capacity additions, at least for major projects. They do tend to ramp gradually as more turbines are added.
Of course there are questions as to how you should treat things like the Rampion wind farm being offline for many months because of cable failure. The same applies to interconnectors, with the fire at Sellindge on IFA1 being one example. The loss of Western Link HVDC for protracted periods on several occasions also will have resulted in much more wind curtailment in Scotland. In reality, we probably need to add back curtailed generation to understand the true levels of surpluses on windy days. The implication would be even more curtailment and/or storable surplus.
All good points. Can curtailed generation be estimated from the payments made or is it not linear? Are the data available?
It is of course one of the huge problems with wind that there will sometimes be too much so we will all have to pay for it to be curtailed. And the more capacity there is, the more that will happen, absent sufficient storage.
Hello Tim,
I suspect that curtailment / constraint payments are non-linear depending on the marginal market in that interval.
Wind proponents say that, because wind is free, it doesn't matter if we have more and more of it. However, all the while that consumers are on the hook for curtailment / constraint payments, this just adds to our bills. A knotty conundrum.
Hello it doesn't... thank you.
Yes, Gridwatch states that wind omits some generation, but it's otherwise is so convenient to use that data and the 'correction factors' derived from DUKES are so consistent that I'm happy to live with the approximations. I am, after all, seeking to understand the big picture not every teeny detail.
I've seen argument on LinkedIn forums by experienced power system people that wind and solar PV act as negative demands. Currently I'm persuaded by that argument.
It then begs the question, which if any of the parameters are 'accurate'? GB doesn't have a massive electricity meter for aggregate demand so that key parameter must be derived somehow. Again, not important for the big picture.
Currently curtailed wind is curtailed for good reasons, either local grid capacity is insufficient or overall surplus generation would unbalance the grid. Power storage able to provide meaningful extra demand would be immense as I've posted numerous times.
Although the arithmetic from the grid point of view adds up if they view embedded generation at the distribution level as negative demand, it can throw up surprises, as it did during the August 2019 blackout, when they discovered that knocking out some demand also knocked out embedded generators, and that the grid disturbances caused embedded generation to trip off line, increasing the requirement for backup generation substantially above what they could see in the control room: they had underestimated the risks. They could get caught the same way if they don't keep an eye on Continental grids and suddenly find several interconnectors at risk of tripping out. It is really much better to model the whole thing properly.
It is important to look at interconnectors too, now they are becoming such a significant slice of the supply/demand picture. I tried to put it all together in this chart:
https://i0.wp.com/wattsupwiththat.com/wp-content/uploads/2023/02/GB-Gen-Price-HQ-1676233783.8236.png
Here you can see the pumped hydro and interconnector flows along with the various sources of generation. Underlying demand is derived from the net export/pumping flows being taken off the overall supply. It becomes clear that we ran extra gas generation to provide exports mainly to France over much of the year, to help them out with their nuclear shortfalls, and as a surrogate way of exporting gas because the gas export lines were full but we could land extra LNG. The role of gas as the first line responder to wind intermittency also becomes clear.
Hello 'It doesn't add up...' - your christening must have been interesting (joking).
As a Chemical Engineer not a Power Engineer I'm interested to use the data - Gridwatch is v convenient - to develop my understanding of the feasibility of 'decarbonisation'. How big the non-fossil sources of energy *are* versus demand, and how the big picture changes if the non-fossil sources are made multiples bigger.
The nitty-gritty of control systems, surge protection, trip robustness, etc. are beyond my area of expertise. [I read the Energy Emergencies Executive Committee (E3C): Final report into the GB power system disruption on 9 August 2019 and understood what it said at a superficial level but not deeply.]
Your chart has the same issue I've encountered with my analyses: it's complex. But yes, I saw that there was a lot of power going *to* France - hopefully GB got paid a good rate for it all.
Hello Tim,
Your comments of 02March arrived while I was on hols, I've only just realised I never responded.
I agree your points 1) & 2), and very much agree with your point 3) as it's the point I was making :)
Re: your point 4) I think I agree with the DUKES approach to estimating load factors; it should get much the same answer with an order or two magnitude less effort. As to the latter part of your comment, we know from the Gridwatch site (if you hover your cursor over that dial) that the displayed wind value is that which is metered, and that there's another ~30% unmetered. I assume that DUKES has methodology to estimate that extra. Them being the official statistician, I would hope so. The differences for solar PV between DUKES, Sheffield Uni and Gridwatch I cannot explain, but the factor is about constant across the available whole years.
Thanks for your efforts, Chris.
In my opinion, DUKES is shambles, and designed to deliberately obfuscate facts about UK energy.
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/1130501/DUKES_2022.pdf
To me, the most annoying fact is BEIS's narrative-based comparisons without providing Tables of Data that would enable its audience to easily compare the different variables and over time.
E.g. Year vs Year; Energy sources; Generation mixes etc, etc.
This is especially annoying because DUKES is 'UK', yet doesn't apportion data England / Scotland / Wales / Northern Ireland. (Or even Britain / Northern Ireland)
We in Britain have our grid & our data, and in real time. Northern Ireland has its SONI data. Naturally, they're not easily comparable.
https://www.soni.ltd.uk/how-the-grid-works/system-information/
/Rant over. ;-)
Thank you, Ron.
I've been kicking myself for making such a basic error with the averages, and you helped cheer me up. But I'm right about that Note 6, it's horrifying if HM Gov policy is being guided using those unrealistic fixed load factors, especially for solar.
The SONI link was also useful, ta.
If you want more accurate capacity factors then you should look at the quarterly data published in Energy Trends.
https://www.gov.uk/government/statistics/energy-trends-section-6-renewables
Not sure if they are starting to over-egg capacity by not accounting for capacity that has been closed or re-powered.
Better still is looking at data for individual wind and solar farms via OFGEM's REGO database. The hazard is it is somewhat in arrears.