Run ASHP / UFH off peak

Hi Everyone,

I’m very interested in thoughts on this.

New build, 2016, assume well insulated house, with wet UFH downstairs. Daikin 16Kw ASHP, no buffer tank, directly connected to UFH and upstairs rads (different zones). LWT = 41C (yes could be lower, but the rads won’t dissipate enough heat at lower temps leading to bad cycling).

Currently, I let the room stats decide when to activate the UFH - using setback and comfort temperatures as required. This obviously means the ASHP runs when it wants.

I have a silly idea to run the ASHP and heat the entire ground floor during off peak hours to ‘pre warm’ the house using cheap electricity, therefore delaying the onset of the stats needing to get the comfort temp during the day. I assume the screed and carpet would hold the heat for quite a while, and ideally, the comfort setting would not kick in until later in the afternoon.
I also have a powerwall, and the idea would be the powerwall would not be drained as quickly due to (a) The ashp not coming on just after off-peak ends to warm the house, (b) the powerwall having charge left in the afternoon when the ASHP calls for heat again (having charged overnight and via solar).

I’m keen to hear thoughts on using the floor as a ‘thermal store’ charging from cheaper electricity.

I know this is not OEM related, but to bend it that way… the UFH is controlled by the 3-relay Harizanov board sold by OEM (with heavy s/w mods) and the whole system monitored by a lot of OEM kit :slight_smile:

For me, this comes down to the balance between a number of factors:

  • It’s definitely good to bias electricity consumption towards lower-cost and lower-carbon times of day
  • With an ASHP, there will typically be a CoP penalty since the outdoor air tends to be colder at the times of day when electricity is cheapest
    • It’s unlikely the CoP penalty will outweigh the tariff saving though - assuming the off-peak tariff is less than half the day-time tariff
  • Once you heat the UFH, that heat is going to come out - and probably quicker than you’d like
    • How quickly this happens will depend on the thermal mass of the floor slab/screed, the insulation (i.e. the carpet) and the room temperature
    • Then there’s the question of how long the house retains this heat for, but if the house insulation is good that should be for many hours
  • Effectively, you’d be predicting a future demand for heat which may or may not occur
    • If you get a lot of passive solar gain, you may find you’ve pre-heated the house only to have it further heated by the sun
    • There must be scope for looking at tomorrow’s weather forecast to influence how much pre-heating to do overnight

It sounds like you’ve currently got a set-back temperature overnight, so a first step would be to get rid of that and aim for the same temperature 24/7. That will naturally do some heating in the off-peak period.

If that works OK, you could try setting the UFH target temp to be slightly higher during the off-peak period and see how that goes.

Big assumption :slight_smile: and consider air tightness - not something mainstream builders care about (except for the reference house which gets measured!). I’d invest in a Thermal Survey and air tightness test - I’d bet there are some really big leaks!

Does this use a blended manifold or does it run at the same temperature as the rads?

I think with Rad/UFH combo, this is a fundamental flaw as the UFH will never want to run at the higher rad temp (unless seriously lacking pipes).

My selfbuild has a deeper than normal concrete slab under the UFH area for just this purpose and IMHO the theory doesn’t stand up to scrutiny in real world applications.

Thanks for the responses.
@borpin UFH doesn’t have a blended manifold, runs at same temp as rest of pipes. I’ve fitted auto-balancing actuators which claim to give a 7C delta. Watching the heatpump F&R temps seem to confirm they are doing this.

Yes, I suspect the property has air leaks, and I don’t think the uPVC is the worlds highest quality either. One thing I did before I moved in was seal under every skirting board → floor as this is a big source of drafts (and also makes carpet edges dirty). I think this made a decent difference as you could really feel cold air coming in there.

Buffer tanks seem contentious, with some people advocating for them and others designing them out. I’ve considered a few times putting one in but have instead tried to match the emitter output to heatpump output by better zoning. I don’t want the upheaval or really want to give up the space needed for one.

@dMb Thanks for these suggestions. I’ve thought long and hard about COP, and ended up with your conclusion. The heatpump may take slightly more power running at night due to temp, but this is vastly out weighed by the cost of on-peak electricity.

I think you’re right on the weather forecast, for both avoiding unnecessary heating and also the likelihood of having PV energy to use. Doing this in practice may be difficult (matching the PV prediction), due to so many variables (temp inside and out, sun brightness, duration etc)

I’ll continue my experiment, tried for first time last night. Sod’s law though, while I haven’t called for heat once today, that’s more likely due to the outside being 15C and sun in the sky rather than floor pre-charging.

If you or anyone has any further thoughts or experiences, please let me know.

I’d also considered the idea of just let the heatpump run at lowest LTW with weather comp 24/7, but the lowest power this thing will consume in any circumstance is 1.3Kw, which would mean around £10 a day.

The problem with the LWT being too low is the rads KW output drops drastically leading to 10 minute cycling on the ASHP as it can’t dissipate approx 5Kw of heat minimum output. Upping the temp to 40C allows the rads to dissipate more at no increased power input (as the ASHP can’t output less heat - unless it’s really cold outside and I’d want higher LWT anyway).

Curious on this statement. Having only ever installed electric UFH in an older house, I’m not sure how wet is installed. With electric, I had to put a layer of insulation under the electric matt to keep the heat escaping downwards(!) and then bury the matt in the screed.

Would it not be the screed that would hold the heat, or with wet UFH is it the the concrete itself ?

If the LWT going to the floor is too hot, I understand the concrete rejects the heat and you get hotspots and other problems. I read somewhere concrete likes to absorb heat slowly. Do you think this would make a difference at 40C ? I’ve no idea how close the pipes are installed (I need to invest in a thermal camera) but can say they are 16mm plastic variety leaving the manifold disappearing under the floor.

The whole system is a bit of a balancing act… just in time for spring when it won’t really be used !

And thanks for the reply, I suspect concrete thermodynamics are pushing the boundaries of this forum.

It is more a conductive layer. Do the calcs and you will see the ability of a thin layer of screed to hold heat beyond an hour or so is very limited. How quickly did the floor cool? That is another guide.

What you are asking (I think) is, heat at 4am and still warmer than surrounding air at 4pm.

What is LWT (London Weekend Television)? :slight_smile:

1 Like

Leaving Water Temperature (Flow temp). I’ve spent too much time in the Daikin manuals of late, I need to go back to speaking english instead of TLA (three letter abbreviation but you knew that one).

1 Like

I’ve got 150mm of concrete here and the UFH pipes are about 100mm below the surface. No floor covering though; the concrete slab is just polished for maximum heat transfer to the room.

Here’s how my floor slab cooled when there was no call for heat for 19 hours (because passive solar gain had pushed the house to 22.5 degrees). The UFH kicked in again around 09:00 on 2023-02-16, hence the blip towards the right of the graph. The readings are from a sensor embedded in a hollow tube buried about half way through the floor slab.

1 Like

Hi @alandpearson, I am doing exactly this since I moved to Octopus’s Intelligent tarrif in December, I’ve got Home Assistant turning my ecodan ASHPs on and off using an automation triggered by average ambient room temperature. The trigger temperature is higher during cheaper rate periods.

I have two ASHPs with one running the UFH and hot water and the other running the radiators.

I’m in a 2016 build which I don’t have heat loss data for. I can say that based on my observations the heat loss appears to be somewhere around 25watts/m2. I’m basing this on the output of the UFH ASHP being restricted to around 5.5kW output whilst servicing 225m2 and keeping the house warm at temperatures down to -10C this winter (based on max 16l/min flow rate).

Octopus Intelligent guarantees cheap rate from 23:30 to 05:30 but in reality often delivers up to 12hrs of cheap rate (when the car is plugged in) so has somewhat skewed my experimentation. Regardless, I have seen that my system does not need to be running anything like 24hrs a day to keep the house warm and is using far less electricity than it does if it is always on.

I have two 11.2kW ASHPs which I believe are oversized by about 200% so results may vary if you have accurately sized ASHPs.


Interesting observations from everyone, thank you.

I’m still experimenting, and as the outdoor temp has dropped a good bit in recent days, it’s made it difficult to understand if I’m winning or not.
In my head it has to better to run 4 hrs off-peak and get the house warm (or not cold) for the morning. So that’s a given. Heatloss throughout the day is harder to quantify as is heating a cold slab as opposed to keeping it warm.

I’ve used a lot less electricity for sure, but I need to do some more tuning as I think the house temp drops approx 1C too much for my comfort.


I have radiators not UFH, so I know it’s different but…

Earlier this winter I’d allowed the temps to drop quite a bit overnight until the heating came on in the morning but that then used a massive slug of energy and basically drained the battery by 10 am.

I changed to heat to 19.5 during the Octopus Go window then drop to 19.0 from 04.30 until 18.00. This uses about 50% of the battery by 10 am and the cheap rate costs are 50-60p a night for the HP so works much better for us.