Earth Notes: 16WW Heat-pump Load Profile

The SSES [govUK2024electricity] government response emerged today, and I was not even half-way through before this pronouncement set me thinking: Manufacturers must ensure that appliances incorporate pre-set default usage hours that are outside of peak hours, defined as 8am to 11am and 4pm to 10pm on weekdays.

What might I meaningfully do with our heat pump settings to nominally comply?

Our morning DHW demand when covered by the heat pump is already in SSES off-peak hours. The evening demand chunk is harder to do much about, especially given the limited number of DHW programme slots in the Daikin.

An analysis of boiler calls for heat from the Radbots via the OpenTRV unit each hour UTC in 2025-01, our coldest month with the heat pump so far:

% gzip -d < data/OpenTRV/pubarchive/remote/202501.json.gz | sh script/analytic/boilerCallFraction.sh
n 8202
boilerOff 4108
boilerOn 4094
boilerOnFraction 0.499147
boilerOnFractionHH 00 0.16954
boilerOnFractionHH 01 0.136232
boilerOnFractionHH 02 0.105882
boilerOnFractionHH 03 0.122449
boilerOnFractionHH 04 0.101156
boilerOnFractionHH 05 0.146552
boilerOnFractionHH 06 0.424855
boilerOnFractionHH 07 0.641618
boilerOnFractionHH 08 0.659701
boilerOnFractionHH 09 0.786127
boilerOnFractionHH 10 0.752161
boilerOnFractionHH 11 0.715152
boilerOnFractionHH 12 0.708709
boilerOnFractionHH 13 0.680233
boilerOnFractionHH 14 0.691429
boilerOnFractionHH 15 0.700599
boilerOnFractionHH 16 0.724551
boilerOnFractionHH 17 0.745562
boilerOnFractionHH 18 0.866667
boilerOnFractionHH 19 0.715976
boilerOnFractionHH 20 0.37092
boilerOnFractionHH 21 0.453521
boilerOnFractionHH 22 0.389381
boilerOnFractionHH 23 0.197015

There are 8202 boiler call samples, about half calling for heat and half not. About 10% of the time a call for heat persists overnight, sometimes because a room did not get to temperature before the night WC setback. In any case these could be particularly cold days that need more heating time. There is only a little apparent extra call for heat beyond that at 05h, before the normal (not set-back) WC curve is in place at 06:10. There is plenty of call for heat at 21h and 22h, which suggests bringing forward the 'normal' start a little, and pushing the night setback back almost to 22:00.

The hours of the SSES morning peak 08:00 to 11:00 have enough heat demand that a small (-1°C to -2°C) WC setback there may be useful, as well as in the existing 19:10 slot.

A -1°C WC setback may result in ~10% heat demand reduction for rooms whose temperatures are being limited by flow temperature rather than TRV.

The existing scheme is:

Something more like this, in terms of setbacks on the WC curve, might work:

• All of the SSES peak period hours have a WC setback.
• The main evening grid demand peak has the maximum (night) WC setback.
• Heating times have been slightly extended to allow for lower-temperature intervals.
• No setback level changes happen on full/half hour boundaries, helping to avoid being part of any 'herding'.

The same schedule would be applied to all days of the week to reduce complexity, and because I understand it to be unhealthy to have too big a difference in routine (etc) between different days.

[brudermueller2025efficiency] suggests that a 1°C setback of the WC curve results in a reduction in energy demand of 2.61%. This implies that a setback of 5°C saves ~13% and 2°C saves ~5%. Some of that 'saving' may be consumed in recharging the thermal capacitance of the house in non-setback periods.

A version with a 1h boost before the main evening setback to try to ensure everything is up to temperature or a little over (though with minimal SSES 'peaks' setbacks since I am sceptical of their value) to better coast through the deep setback:

GB carbon intensities by hour available from/with:

curl -L -s https://api.neso.energy/dataset/88313ae5-94e4-4ddc-a790-593554d8c6b9/resource/f93d1835-75bc-43e5-84ad-12472b180a98/download/df_fuel_ckan.csv

There is an argument to put back by 1h the post-peak DHW 'eco' slot, primarily there to cover the occasional evening shower or bath. Yesterday and today (2025-06-28) GB demand did not peak until ~19:00Z, ie 20:00 local. Though currently all heat-pump DHW demand is covered by the Enphase batteries.

The current scheme is:

The new scheme is:

I might drop in the change ~2025-07-01 to see if anyone notices. Our DHW usage pattern does not change hugely with season. We might come unstuck when Ellie is home in winter and she and Morgan both want an evening shower/bath, since 1h20 is not really enough time to guarantee a full warm-up, especially while I am limiting heat pump power in.

(I see that DHW top up in the evening has not happened on many/most recent days, I assume because the tank losses are smaller with the ~25°C kitchen temperature.)

The Daikin only allows six heating-schedule weather-compensation curve-adjustment events per day, so the pre-evening-peak boost is going to have to go. I am putting this schedule in to the system this evening, ready for December!

A trial by Ovo/Nesta [wollard2025flexibility] using a 2h/4h +1K setpoint preheat and a 4pm–7pm -1K setback claims to have shifted approximately 30% (~0.75kWh/d) of heating demand away from the peak period.

On enquiring yesterday about water temperatures I was told it was not warm enough some times. Though first thing in the morning seems to be OK, washing up after the evening meal may be more of an issue. I might be able to tackle that by reducing the hysteresis or scrapping the evening eco slot from 20:10 and moving an elevated temperature slot before 16:00. As a small experiment I have this morning raised the initial main block from eco (45°C) to comfort (48°C). There will be an energy cost, from lower CoP and increased tank losses.

This could be automated to some extent by using DHW weather compensation, to keep the DHW tank cooler in summer and reduce standing losses, when diverted PV can be expected to provide hotter water. Greater losses from the tank to the kitchen in winter are less of a nuisance and would indeed partly substitute for space heating. A two-point WC curve with an external temperature range between (say) 10°C and 15°C (the latter is the space-heat activation threshold), corresponding to DHW target of 48°C to 45°C (or lower, down to 42°C) might be reasonable, possibly with ~24h external temperature smoothing.

Initial data suggests that this may be costing an extra ~0.15kWh electricity (from ~0.45kWh to ~0.6kWh) in the morning 6am DHW heating slot.

Octopus offers a Cosy tariff for heat pumps (for homes without an EV):

• Three cosy periods of super cheap rates between 04:00 – 07:00, 13:00 – 16:00 and 22:00 – 00:00 every day, 51% cheaper than the Day rate in your region.
• A peak rate between 16:00 – 19:00, 50% above the Day rate in your region.

That corresponds to 7 (seven) pricing periods per day. My Daikin only allows 4 DHW and 6 SH changes per day. And we all tend to go to bed at 10pm or soon after, so that last 22:00–00:00 slot would be difficult to use with the heat pump, though could be used for charging the (Enphase) battery (or Thermino), or running the dishwasher.

The electricity that we use at peak rate, mainly cooking, would need considering.

Thought experiment: if on this tariff, how might things be organised to minimise cost and carbon?

Here is a plan:

• Try to minimise consumption at peak, if possible, maybe shift some cooking earlier.
• Run the dishwasher in one of the cheap slots if not covered by solar during the day, preferably 04:00–07:00 for grid friendliness.
• Ensure that the battery has enough charge to cover peak period demand, force charging from grid in the 13:00–16:00 slot if not already full enough from PV, so especially in winter. Possibly extend this force charging to each of the cheap slots to (just) cover anticipated demand in the following non-cheap period.
• Adjust the morning 'green' period for pasteurisation to use the 04:00–07:00 slot.
• Have DHW set for comfort in the 04:00–07:00 and 13:00–16:00 slots (expecting the second one to force a boost), eco between those two, and reheat the rest of the time, ie from 16:00 and overnight.
• Have space heat run WC+1K in the 04:00–07:00 and 13:00–16:00 slots, WC-1K between them (to cover the SSES morning peak), WC-5K in the 16:00–19:00 peak slot, then WC-1K from then until 22:00 to cover the SSES evening peak, and then maybe WC-5K overnight until 04:00.

All of this upheaval might save ~£50 per month for 3 or 4 winter months, thus maybe 30% of our predicted ~£600 annual energy bill. No behaviour change is needed if the battery can be force-charged when needed in the 13:00–16:00 slot to be able to carry all the load in the following peak. That saves money but slightly raises carbon emissions from round-trip losses. (Note that the battery cannot meet all peak power demands from cooking, nor from the heat pump when running without a power limit eg when defrosting.)

Looking at a sample of hourly demand as seen by the Eddi for the ~20 first days of this very mild December for the evening peak hours (hour UTC, samples/days, hph4 W, grid W) suggests ~2kWh to cover all house demand during this time as-is:

16:00:00,20,69,311
17:00:00,20,198,1135
18:00:00,19,138,484

That would take under 2h of the cheap slot to load into the battery from grid.

Looking at consumption as seen by the Enphase 2025-01 (15-min UTC block, samples/days, gross consumption W, grid W) which has been the coldest month so far with the heat-pump, suggests that a mean of ~2.5kWh would be needed in the battery, still feasible:

16:00:00,31,439,364
16:15:00,31,521,460
16:30:00,31,690,647
16:45:00,31,844,808
17:00:00,31,1032,1005
17:15:00,31,1417,1362
17:30:00,31,1457,1439
17:45:00,31,935,937
18:00:00,31,751,755
18:15:00,31,687,676
18:30:00,31,650,646
18:45:00,31,702,697

The heat pump has three levels of 'quiet' mode ('quiet', 'more quiet', 'most quiet') which seem to cap maximum output to 75%, 50% and 25% of maximum. 'Most quiet' seems to be recommended against in cold weather, and might indeed reduce output below our design-day ~2kW demand.

Today (I am at home, there is low (de)frost risk), I have experimentally put the pump in manual 'quiet' mode to see if it blunts the DHW demand peaks, maybe getting it to within the Enphase power ~5A power limit.

There is a possibility of this reducing CoP, though reports on line suggest that this is in fact not an issue.

It may be useful to set a 'quiet' schedule to cover DHW generation and the SSES and grid peaks. That might be 'quiet' from ~06:00–10:59, 'more quiet' from ~16:00–18:59, and back to 'quiet' from ~19:00–21:59. Or possibly 'more quiet' for all of these times to get closer to 5A max. Or '(more) quiet' from 16:00–10:59, to also reduce any night-time noise.

Noise from the external unit has simply not been an issue at all so far, to be clear.

The evening DHW run starting at 20:10 (outside is ~8°C) seems to be showing a peak demand of ~1800W, higher than the Enphase could cover at best. So plain 'quiet' mode would not be enough to avoid drawing from grid when the heat pump runs when there has been recent PV generation, at least at in this chilly weather. It might be enough in warmer weather when the DHW CoP is higher. This evening's DHW run does seem to have been wider and lower, and possibly using a little more energy in total, than yesterday's.

After the DHW run finished at ~20:50, as further experiment I put the heat pump into (manual) 'more quiet' mode to try to moderate the next morning run's peak demand. The setting probably cannot be left there if the house and DHW are to come up to temperature in reasonable time on very cold mornings, though it would be unlikely to cause outright failure.

2026-01-23

The 06:20 morning run took ~25 minutes (outside is ~7°C) with imports peaking at ~1800W, so maybe a fraction lower than last night, and consumption maybe slightly higher, but nothing dramatic.

I will try a 'most quiet' setting before my bath, and DHW replenishment for it.

My inclination is to go for a 'quiet' schedule from ~16:00–10:59 covering DHW runs and SSES peak times and night-time, but fully unleashed during daytime when needed for very cold days.

~10:20 'most quiet' mode engaged, while space heating running. What looks like the following DHW cycle at ~11:00 for ~35m seems to have drawn ~1000W–1300W (full imports ~1250–1500W), so could be within the Enphase power limit.

I am setting a 'quiet' 15:50–11:09 schedule. I hope that as weather and CoP improve then the peak demands will simultaneously shrink to fit the battery power limit... Meanwhile, DHW demand should be a bit less spiky.

Under this new regime, the evening DHW run had grid imports peak at ~1900W, which implies peak heat pump demand at ~1800W or less, much as expected. Two nights ago the imp/hph4 was ~2600W/2500W. That morning the imp/hph4 was ~3300W/3200W.