Posted by: @majordennisbloodnok

↑

Nor can one extrapolate how green the power is from the unit price.

The price is still a pretty good guide, I'd have thought. Agile peaks from 4-7pm every day and that is surely because the demand that spikes then can only be met by firing up the fossil fuel power stations. If you are on Agile, the biggest saving you can make is to time-shift any usage that can be moved to any of the other 21 hours of the day. The carbon intensity of those other 21 hours is surely going to be much lower than those three peak hours.

 

You’da thunk, wouldn’t you? But that’s not borne out by the data. For instance, on 2nd September the peak carbon intensity (300g/kWh) was at 9am (about 21p/kWh) and 4pm (36p/kWh) was only just over half the peak intensity figure (about 160g/kWh).

Despite logic suggesting it should be, historically you find price is not a good proxy for carbon intensity.

Posted by: @majordennisbloodnok

↑

But that’s not borne out by the data. For instance, on 2nd September the peak carbon intensity (300g/kWh) was at 9am (about 21p/kWh) and 4pm (36p/kWh) was only just over half the peak intensity figure (about 160g/kWh).

Despite logic suggesting it should be, historically you find price is not a good proxy for carbon intensity.

 

You may be right, but is that one day just a one-off or is it representative of the norm? If there are figures to show that carbon intensity normally peaks other than from 4-7, then I'd agree that price is not a proxy for carbon.

 

No, 2nd was not a one-off. No, the carbon intensity doesn’t normally peak at another time. In fact it doesn’t normally peak at any particular time.

One of the good things about Home Assistant is that I can look back at past data. Looking at the historic Agile prices and the historic carbon intensity over the last month, they appear to my amateur eyes to be independent of each other; sometimes coinciding, sometimes conflicting, sometimes obviously unlinked. I’m not a statistician so can’t say categorically there’s no link but if there is one it certainly isn’t obvious. 

@ivanopinion

This was interesting and relevant 

https://guylipman.medium.com/power-price-vs-carbon-intensity-d97ee6a70aaa

You’re right, @bontwoody. That is very interesting.

@majordennisbloodnok Think I might download some data and see how much statistics I can remember

Posted by: @bontwoody

↑

@majordennisbloodnok Think I might download some data and see how much statistics I can remember

You're a better man than I am, Gunga Din.

I work with technology all the time and my general maths and logic skills are perfectly up to snuff. My statistical analysis skills, however, are not. I have a statistician and a physicist in the family and as soon as they start talking shop I find myself lost.

Nonetheless, if you need historic data I can pull some off for you. Not sure I've been following the carbon intensity for a full year but you're welcome to what I have if you have any problems getting it online.

 

@majordennisbloodnok Thanks. I’ll happily take you up on that to save the need of running the python script. I’m in the middle of reading a statistics book so it’s a good opportunity to reinforce what I’ve read.

@ivanopinion @majordennisbloodnok 

Managed to have a look at the data today, unfortunate the data you sent me had different timestamps not in the same intervals and it was not trivial to try and clean it up.

I went back to the horses mouth via the api and finally managed to get some data I could analyse for 11/09/2024. Obviously this is only one day and I will get a bigger sample when I have more time but on that day the correlation between Carbon Intensity and Agile Price was 0.8 (Pearsons coefficient) so pretty good.

It may be that there is something predictable that affects the correlation, such as the amount of green energy in the system. im interested enough to carry on looking

 

@ivanopinion @majordennisbloodnok 

So I managed to get a bigger slice of data from the 17/08/24 to 13/09/24 and the results are interesting. Overall the correlation between price and carbon intensity was 0.58, so a moderate correlation, but this hides very big swings with 9 days having a correlation over 0.8 and one day having a very small negative correlation. The summary data is below if you want to look at it.

 

Thanks, @bontwoody; that's very interesting and useful. It does, however, leave me in a quandary.

You have demonstrated that there does indeed seem to be some correlation between price and CO2 intensity contrary to the conclusion I came to from looking at the graphs. In other words, I was wrong. You have also demonstrated - and informed us - the correlation is not huge and hides big swings. That gets me very unsure again as to whether price is a reasonable proxy for carbon intensity (because there is a correlation) or a bad proxy (because the variations are large); there's the quandary. 

Eventually, I'm erring back to the pretty obvious point that if I can receive accurate carbon intensity figures then I don't need a proxy for them, and that making any decisions based on both an accurate price and an accurate carbon intensity should be a better fit for my purposes than decisions based on just one or the other. I'm mindful, though, that I want to avoid cherry-picking what you've uncovered to suggest it supports what I've been saying and to avoid throwing out wholesale other reasonable interpretations of what you've found. As ever, honest objectivity isn't always easy.