Running the following data through claude to quickly calculate and visualise the median and mean unit rates that HeatpumpMonitor.org heat pumps would achieve on different time of use tariffs (assuming no change to their current running pattern of course).
Useful for back of the envelope spark gap calculations, e.g with gas at 7.3 p/kWh, the median spark gap for heat pump monitor systems if they were running on agile would be 2.41.
As I understand it the mean e-SCOP equivalent to the mean SCOP 4.0 on 26 p/kWh price cap for these would then work out to be 5.9.
Running the same on all systems if they were running on Octopus Cosy, showing that in general if most heat pumps shifted to cosy they would be worse off than agile but there are clearly some outlier systems that are achieving maximum optimisation on their chosen tariffs E.g:
Important caveat that owners would do better if they shifted the load into the cheap periods, even though this is not optimal running for a heat pump.
Add a modest battery, and you can effectively run all day at the cheap rate, charging up 3 times a day. I averaged around 15p/kWh last winter using a combined strategy on Cosy, beating the average rate for the previous winter on Agile (it was unusually expensive then).
Agreed, and we can see the owners that have done this in the distribution. It’s probably a fairly safe bet to say the median cost reflects no time shifting, the minimum’s of ~13.5 p/kWh are close to the maximum achievable optimum (without a battery) and 15-16 p/kWh on agile and maybe 18-19 p/kWh on cosy is probably closer to normal optimisation? assuming only a few users are shifting their consumption and that part of the tail reflects the gains made..
My system is not on heatpumpmonitor, but I can confirm that the winter in which I ran my system on Cosy with load shifting and no battery, I was able to average 19-20p import for the winter.
With a battery, we now only import at Cosy cheap rate, recharging 3 times per day.
Hello @BenY there’s no field to enter the tariff your on now. What we do instead is use the half hourly electricity consumption of each system and see what it would cost to run that system on the different tariff options; agile, go, cosy, e-on next pumpted etc.
The owners that have optimised to one of these tariffs will likely come out cheaper on the tariff than the median which will more likely reflect more normal comfort / always on running strategies.
I’m on cosy tarrif because it lets me charge the battery multiple times, which coveres all my winter heat pump use with cheap slots. On GO I would run out and be charged peak. GO is also close enough in price (9p) that I couldn’t be bothered to optimise when to switch from 12p during the summer, also you run the risk of the fixed price changing when you need to update it…
This is all despite my EV using probably being the most energy used every day and that could be booked into GO slots, it’s just not worth to risk paying for any peak energy.
The same thinking also stops me getting any more battery storage, what’s the point to save 3p per unit… Would take forever to recover costs.
This is all massively too complicated for your average Joe. The more you think about it fixed and variable pricing is by itself already too complicated.
It would be much better if we had no standing charge and replaced it with something like “standing size”, then grannies with almost no energy use would get rid of a standing charge penalty and people with batteries could trickle feed under their supply limit… Seems like that would be healthier for the grid.
I feel like I still can’t quite wrap my head around this properly!
My gut feeling is that agile is sneaky, because what really matters is the price difference between just before the peak, it’s exactly when you would need to top up your battery to cover you for the peak. I don’t think this cost is tracked anywhere.
A classic example of a sneaky day would be 4p overnight (great), 18p during the day (worse than cosy) and 32p on the peak… You can see how your merry battery charge would be depleted just as the peak arrives and you are forced to take 18p energy.
Hopefully I am describing this properly…
It’s certainly easier on the brains by taking cosy and charging my battery 3 x times a day … Being way cheaper than anything else. Unless GO and battery could cover my whole winter use (it can’t).
Ultimately this feels like shuffling chairs around the issue of energy being so stupidly expensive.
I ran on Agile for a year, and developed an algorithm to intelligently manage charging and discharging by looking ahead at prices and solar, etc. It worked well, but would suffer on those few expensive days (weeks).
I switched to Cosy, which is much easier to manage thanks to the fixed daily prices and is overall more predictable.
Depending on your battery manufacturer, some (many) now have AI and are capable of importing the Agile tariff, so should, in theory, be able to calculate when to optimally charge on any given day knowing previous usage patterns. At least in theory - not sure how well that works in practice.
Of course what really matters is the longer term cost average rather than any individual day and whether you occasionally run out of battery and need to import 5kWh at peak rate. I know from previous experience that my cost average was ~19-20p per kWh when running Cosy without a battery and that reduced to 13p per kWh (or whatever Cosy cheap rate currently is) with the addition of a suitably sized battery. So that is my known baseline when considering if switching to a cheap overnight EV tariff (or Agile) and/or adding more battery storage may be financially beneficial.
As noted above, comparison apps like OctopusCompare are of limited usefulness as one’s import pattern on a tariff like Cosy (or GO) will look very different to a strategy that is optimised for Agile.
Ultimately my human error - I thought the script that I wrote a while back to calculate the rates was still fine but on closer inspection it wasnt so! The agile tariff feed I was pulling the time series agile tariff data from stopped returning data from the Octopus API on the 28th of March of this year and I didn’t have a guard for null values and so the tariff rates were parsed to zeros resulting in incorrectly low average unit rates for the full year. The Cosy rates were also based on an earlier version rather than the variable changes over time
I’ve just re-run all the numbers and also updated the implementation of the cosy and go tariffs to use the API based data rather than manually entered prices.
Interestingly the gap between Agile and Cosy has actually widened from 4.7 p/kWh to 5.7 p/kWh and cosy without flexing is basically very close to the new July price cap.
The median Agile e-SCOP is now 5.1 not 5.9 and the spark gap is 2.8x not 2.4x.
