ASHP sizing for Freezing Conditions

I’d be interested to know why. I’ve invested in a wifi switch for one of the 3kW immersions in the UFH buffer tank which I intend to turn on for 4 hours on Octopus Go overnight to supplement the ASHP on cold nights.

Like you, I see a significant drop in the ability of our ASHP while it’s having to defrost. So my thought was to supplement with the immersion in the UFH tank.

We have a 9kW ASHP which feeds the UfH in an insulated slab. When the weather is around 10C, then a 4 hour run on the off peak Octopus Go tariff is enough to keep the house very warm (shorts and t shirts warm), so our usage is about 4 hours at 3kW @ 9p per kWh.

But when it’s cooler, the ASHP cycles through heating and defrosting and the UFH buffer tank temperature doesn’t really increase like it does on warmer days. So it’s a bit cooler - even have to put a jumper on…

We’ve only just set this up, so it will be interesting to see what the actual effect of the additional 3kW into the UFH buffer tank will be - but I can’t imagine it won’t make up the deficit (as per the @ColinS’s statement above). I should add that we only use the ASHP for the UFH, so there’s no contention with heating or hot water.


If you get a COP of > 1 heating the buffer tank then that’s more efficient than running an immersion heater.

It is indeed - the immersion will be in addition to the ASHP - but only on days where the outside temperature drops to say 5C. So instead of 9kW at COP 3, it will be 12kW at an effective COP 2. COP 3 for the ASHP for 9kW and COP 1 for 3kW.


I didnt have all of my convector rads installed when the first cold spell hit…early December 2022. The lack of impact of the back up heater then might have been due to radiator limitation. I’l be able to see if this us true when we get another cold spell.

And please excuse me when i confuse back up and booster heaters. I just wish they would use immersion for the hw tabk heater to avoid this.

1 Like

Looks like there’s another cold spell on the way, so we should be able to see what happens.


With frost outside, I report on a very different type of 12 year old heat pump, which heats to a blower on the wall in a single room, and is usually called an ‘air conditioner’. I got this before feed in tariff and CO_2 efforts became mainstream, at a time when ‘coefficient of performance’ was not a well known phrase. It uses 0.7kW of electricity for typically half of the time, and blows “more than” a kW of warmed air.

It circulates refrigerant R410a, which is not the best choice for heat collection, especially in cold weather. Yesterday while T=9C outside, performance was close to CoP= ‘about 2’ (there is nothing to measure or log it. I got that from direct comparison, alternately heat pump for an hour, OFF for an hour, plain resistive electric heater for 1/2 hour … )
That is, if your measure of usefullness is pennies per kWh at usual residential energy prices, the R410a heat pump was going to cost more per kWh that plain old gas heating in cool weather; 9C, and would be worse today now that weather is freezing.

I have solar panels on the roof which frequently generate a net surplus of more than 0.7kW, which is enough to run that R410a air conditioner heat pump completely free of energy cost. If the sun gets higher today, I’ll switch it on in addition to (not instead of) other heating, expecting even lower CoP. The decision is “0.5 kW plain electrical heater, or 1kW fan heater, or 0.7kW to the heat pump?” while solar power is available. I’ll be using the normally-off gas central heat somewhat; this small heat pump is only about getting additional free heat to only one room.

Do you lot at Emon want the results of cold day testing of this R410a air conditioner ?

1 Like

@Stephen_Crown Good spot! I’ll disable the 2C uplift setting on mine, see if that helps to reduce defrost cycles.

1 Like

I was hoping my Daikin would perform better this Winter with a full suite of well sized radiators and set up sorted out but I was disabused yesterday when it cycled as rapidly as it has ever done. … and just at zero c and not much lower. Humidity was not that high - around 80% so in line with the standards - seemingly destroying my theory that saturated air was the cause of the problem. The unit was driven to it’s limit of output as my house was warming up from 11c after we had been away for the week. I’m still only getting 7.5kw max when it goes into rapid defrost mode - you can see from the above how the defrost periods increase as the outside temp climbs in the morning.

The bottom line is that my ‘10.6’ kw model can run with a COP of up to 5 at 10c - producing 4.5kw without cycling but that can only increase to 7.5kw due to defrost cycling at zero c. Not much of a useful range when it is not cycling one way or another. The defrost is obviously needed when I look at the snowball created at the end of the 30 mins… so it is not over cycling as far as I can see.

I see that another 9kw Altherma near to me in Farnborough had a similar cycling issue albeit with longer 1 hr periods and a lower output - presumably it wasn’t being pushed as hard as mine.

And another nearby in Fleet had the same fast cycle time as mine.

The Cambridgeshire Grant Aerona is doing something similar

As is the 7kw Vaillant in Rushden

Once again this suggests to me that if you calculate that you need an output of X at say -2c then you need to buy a heat pump with a stated capacity of about 1.33 X in order to really produce what you need it to produce in freezing conditions.


@ColinS you’ve described exactly how mine works to a T.

Thank you for this as it confirms it’s not just me.

OpenEnergyMonitor (thank you @TrystanLea & @glyn.hudson & @Timbones plus others who contribute time, code or otherwise) has shown me two things:

  1. It appears to not want to be driven hard when cold and humid as it just destroys CoP and ultimately net heat energy output given the crazy cycles frequency, which is a vicious circle as it then has to work really hard to come back from it. Kills the idea dead really for tariffs such as Cosy Octopus IMO. (Ironic these both are Octopus :octopus: product offerings!)
    I have reinstated the Honeywell controls for the UFH loops (~500m for ~40m2), and am trialling turning off 300m of UFH for extended periods to see what effect that has)

  2. Most efficient running appears to be tracking the minimum output of the unit by nudging target temperature (can the “target temperature” be added to my heat pump app?) to avoid pushing it beyond minimum, yet stay within the maximum 4C overshoot window. (Big caveat on this one: radiator/emitter sizing, water volume and heat loss need to be high/large enough to cope with what it wants to get rid of at minimum modulation)

Here is my data on HeatPumpMonitor (via ESPAltherma, myenergi data)


Yes, you don’t want to undersize your heat pump!

I carefully designed my system to run with a 35C flow temperature at -3C and chose a heat pump with slightly more than the calculated heat loss of the house. It performs very well at over 6C, but we’ve had a few days recently with high humidity and low temperatures and the heat pump doesn’t really cope. It struggles to get up to the desired flow temperature. The issue is probably that the heat load of the radiators is greater than the capacity of the heat pump at low air temperatures. I’ve turned a couple of radiators off which lets the flow temperature rise, but it still runs continuously.

As it’s running at full output most of the time it freezes up regularly and the defrosts kill the CoP. It had been giving a CoP of over 4 but in theses conditions it drops to under 3.

To be fair, at 0C it is still keeping the house at the design temperature and the CoP matches the spec (2.67 at -2 or 2.98 at +2 with 35C flow temperature). But I should have gone up a size on the heat

On this day it starts happily enough as the outside temp drops from 10 to 6 then it starts struggling as the temperature drops to -1.


Thanks @ColinS

This seems like a very plausible multiplier. I have somewhere between 1.4-1.7x on mine. Accurate heat loss somewhere between 3-3.5kW, heatpump rated output around -3C: ~4.8-5kW.


One thing I want to be sure about here is that I think both your Daikin @ColinS, the Farnborough example and the vaillant are all showing the compressor easing off early in the cycle suggesting target flow temps are being reached, and so not necessarily testing the maximum output. It’s worth setting the target flow temp out of range e.g 50-55C to see how hard the heat pump will really try to run. The compressor should try and max out for the duration of the cycle before going into defrost I think…

1 Like

More by luck than judgement I seem to have a heat pump that is quite happy in these conditions, because it is in theory too big for the house.

Despite my ranting I had no intention of changing it because I thought I would be glad to have it when the conditions are right and I have no confidence that the smaller Daikin heat pump would work well enough when I really needed it.

I would prefer to adapt my house in terms of radiator sizing and heat loss to suit the heat pump when it is warmer than compromise the cold weather performance or capability.

Not very scientific but looking at other heat pump data it seems that pushing to the upper limits when it is freezing conditions is an issue.

Being able to operate under the upper limit seems to be advantageous.

@ColinS was right as usual, thanks Colin.

1 Like

@matt-drummer if your heat loss is 4.5kW then a conservative 1.5x multiplier would suggest that you need a 6.75 kW heat pump, perhaps the 7kW Daikin would be fine for you? It would be worth collecting data through at least a couple of decent cold spells first


(I’ve added this to mine, but thought it was a bit niche)
(Temporary workaround is point heatpump_roomT at target temp feed)


Hi Trystan,

I am really happy with this heat pump over a cold weekend, it’s working pretty well.

Because I can, my solution is to heat another area when I can and turn it off if I need to for the main house, sounds like a posh estate but it’s not!

I have my timber office I can heat, it will solve my excess heat when it’s warm and I can turn it off there when I need it for the house.

I don’t want to change my heat pump, I can potentially see times when a smaller one would struggle despite my low heat loss, they only need to lose a bit of performance and then defrost a lot because they are being pushed to the limit and you are then struggling for heat.

I couldn’t swap my 9kW for and 8kW, it’s a complete unknown and would be a leap of faith that I don’t have.

What I lose on the swing I seem to make up on the roundabout.

I think there is a big problem with heat pump design.

My experience is only with this Daikin model but they all seem to have a limited range of operation.

That leads us not to choose a big enough heat pump to comfortably cope with the coldest weather we might encounter because they then perform badly when it is milder.

What we need is a heat pump that works efficiently over a broad range of weather conditions, I assume that would be a compressor with a greater range of modulation?

There are good installers and not so good installers but it seems so critical to get everything right, there is very little margin for error if you want decent performance in most conditions.

I can understand why they tend to oversize, despite my disappointment with my milder weather performance that is nothing compared to how unhappy I would be if I was unable to heat my home when I really needed it.

I think we would all be happy to oversize if we thought the heat pump would be ok when it’s not so cold but that’s not how it appears to be.

I think an overlooked point on the “oversizing” topic is “system design”.

And I’m not talking about the heat pump, pipework and radiators “system” in the home.

Zoom out, and the wider system is the distribution electricity network (and wider still, the transmission network), including batteries, generators etc (which can occur on both these days)

Sizing to the limit (or under as I’ve seen some people on social media suggest (!) - say 0C-2C) to optimise for milder weather CoP, then compounding defrost net heat output deficits leads what some want, or what will then occur, for backup heaters to come on (remember at a CoP of 1.0) to make up the deficit, just as everyone else will be doing the same.

Can you imagine if all of a sudden in a cold humid snap, heat pump efficiency dropped off a cliff, leading to what would in aggregate be quite a severe shock to the “system”.

I deal with system design as part of my work, and one of our points we keep having to reiterate to others is:

“what part of, or which system are you optimising for?”

Because that optimisation (good or bad) has an effect on other parts of “the system”.

There are solutions other than sizing the primary A2W heat pump for the design condition.

For example @marko excellent idea of sizing A2W to ~say 2C rather -3C, and then opting for a minisplit A2A heatpump that can serve double duty: cooling in the summer and picking up the additional load in the worst of winter. This option for example preserves wider system efficiency (as both are heat pumps). A hybrid system however, so some end user control downsides.


My Daikin, the older gen 11kw Altherma in Farnham has performed quite well over this cold snap.

Following advice from @ColinS a few weeks ago I changed to a fixed leaving water temp of 37C. I have been getting run times of around 2 hours, in the milder weather and over the last few days too. COP has dropped a bit as it’s got colder but it’s not tanked like some of the other Daikins. The pump has actually turned off in the evenings because the house has got too warm after running continuously.

Thought I’d share one that’s behaved ok.


Yours is oversized too.

Oversized or maybe I should say correctly sized(?) heat pumps perform better in freezing conditions it would appear.