ASHP sizing for Freezing Conditions

I think this issue is common to alI Air Source makes so would value input from anyone with experience and understanding of what happens with heat pumps in Freezing Conditions and if there is any way to set them to minimize the problem. I don’t think it matters too much - for this issue, whether your emitters are radiators or underfloor, mine is a mix of radiators and convector rads. I have enough emitter capacity to run it at 38c and get the required output at -2c outside.

I’ll start by with my story over last Winter and then go into my understanding of the issues freezing conditions raise as to Heat Pump Sizing.

In late 2021 the local Daikin rep was the only one of 6 people I first contacted to respond, hopefully it is different now. The main issue I felt I had was correct sizing. I gave them a detailed MCS analysis. I calculated my house heat input need at 9.8 kw at -2. We had a very cold sub zero week that Winter so I range-rated my gas boiler to a max output of 11kw and it easily kept the house warm - so empirical confirmation of my theoretical work. 9.8kw is the requirement!

With this data Daikin still wanted me to have a 16kw machine with a low loss header. I talked to two real experts and they both said to go for the minimum size to match my design temperature requirements and no low loss header….so I went for the EDLA 11 = 10.6 kW unit. Daikin also insisted on a Volumiser of 20l – again crazy as my system volume is 120litres, but I gave way on that.

As a final check I asked the sales rep for tech details on the performance at -2c and he showed, but would not give me, a picture from the tech data book. It isn’t generally available but the Daikin forum has a copy which I now have. It shows that the 10.6kW will give close to 9.8kw at -2c – so I was confirmed in my choice.


What I now realise is that foggy, damp British Winters usually have a saturated or even supersaturated air at 0c whereas the test data is all done with dry bulb/wet bulb differences of 1 degree. The tests listed are all standards by which I guess all manufacturers have to comply. At 4c and -2c they equate to roughly 80% relative humidity. With this realisation I recently looked again at another table in the Daikin Tech Data Book.

Notice something odd about all four Daikin Altherma model outputs in the second listed BAFA test at 2/1 and 35/30 ? What on earth is this test?

I now understand that the humidity difference makes a huge difference in the frosting up of the unit – at saturation mine has to defrost every 35 mins for 6 mins. At 80% RH that becomes every 3 hours for 6 mins. Which explains why I only get out an average 7.5kw at -2c and flow temp of 38c. I have to supplement the heat in my house for a few days/weeks of the year. The 3kw booster heater should make up teh deficit but somehow does not seem to make any difference at all - curious? So my unit does not perform at all well below zero. Given that the evaporator size is the same with the 11kw and 16kw unit I very much doubt that the ‘bigger’ unit, which with this ‘range’ from Daikin is not really any bigger, would give any more. Having said that my heat pump is working great above freezing and I’m generally very pleased with it.

I am not sure that the Daikin tech guys really understand this – or Octopus, but I suspect that they over size to avoid the problem. Has anyone found more understanding amongst other suppliers?


Thought I’d add my more detailed calculations on this topic. First the difference of 1 degree between dry and wet bulb - shown in all of the tests in the table in my first post.
WhatsApp Image 2023-11-08 at 10.03.26_28133a6d
For -2/-3 the RH used in the MCS test is 79%

Now use this splendid web tool to work out the grams of water/m3 air these RH represent.

Humidity calculations using temperature, relative humidity, pressure (

I’m assuming that the air in contact with the evaporator comes away at a maximum RH of 100% at a minimum of the evaporator temp. It won’t be 100 % efficient but this is a first pass assumption. So if we take the Air-In grams per metre and assume a temperature drop of say 4c then the difference to the Air-Out grams per metre at the lower temp should give us an idea of the ice deposited.

-2c RH 79% = 3.33 g/m3
-6c RH 100% = 3.17

Difference in test data = 0.16 g/m3

Now with saturated air in, -2c RH 100% = 4.22
Difference to -6c in foggy freezing conditions = 1.05

1.05/0.16 = 6.6 times the ice build up or the difference between defrosting once every 35 minutes and every 3 to 4 hours!! Is this correct? If so it just about explains my capacity loss. The heat pump goes into reverse during defrost taking heat put of the radiator water.

So in my case it is taking 2.4kw out over 6 mins after putting 9.5kw in over 30mins. Average = 7.5kw

In the MCS test it would take 2.5kw out over 6 mins after putting in 9.5kw over 198 mins. Average = 9.1kw


Central continental climates generally have very cold and very dry winters so will not see such a problem. Is it only in warmish damp Britain, a maritime climate, that we have the problem?

Incidentally my defrosting starts at +4c outside . The test data vs Foggy reality defrost delay is about 3 times in those circumstances.

And finally here was my struggling Heat Pump on a cold night February 2023.

The average output during cycling was 7kw.

If the coming Winter will be your first - be prepared!


I would say it doesn’t perform well around zero.

When it really gets cold out (below zero) then you get snow and this defrosting headache largely goes away.

1 Like

Correct - I doubt there would be a problem with cold dry easterly winds here.

Some of that will be actual defrost.

Some of that will be pishing heat into the outside air because the defrost algorithm is incredibly conservative.

It does look like this particular unit is trigger happy on the defrost.

Completely agree that “peak capacity loss due to defrost” and “design condition” overlap unhelpfully in the UK climate. You don’t notice it so much in the continental climate because you don’t need the capacity at the time it is most impacted.

“British standard air” genuinely != “Euro standard air” in this case.

I disagree that the heat pump was struggling on that night in question. You were emitter limited rather than heat pump limited over the period shown. It’s probably fair to say that the unit doesn’t appear to compensate for defrost by bumping up the target flow temperature to compensate though.

Nice writeup BTW!

1 Like

On the morning shown , and others when it was very cold, the unit really could not get the flow temp above 39c before the next defrost started. So the average flow temp was below the 38c needed by the radiators. Putting the flow temp set point up a few degrees wouldn’t have changed this so, while I half agree that the radiators might have been limiting (the room temperature increase) they were doing so because the heat pump could not reach the flow temperature.

1 Like

The other thing I noticed today is that the normal refrigerant temp in the evaporator is 4c below ambient. When the unit is working easily. During the defrost battle though you can see it is dropping to 10c below ambient, presumably as the ice blocks the evaporator it tries to keep the heat input high by this method - which must become a vicious spiral as the lower temp promotes more ice…

I might dry a lower flow temp target this Winter and a lower set back temp overnight so it does not have so much work to do in the morning…

Thank you for posting this explanation @ColinS!

No problem - just had another thought though - shouldn’t the latent heat of freezing boost the effective evaporator capacity around zero - at least to start with?

The original octopus energy heat pump did just that.

Crazy dark winter Swedes. Direct expansion system with a few kg of propane in them. Big sticks in the garden that would freeze (to take heat from moisture) with defrost instructions somwhere online of whack it with a wooden stick if the ice won’t fall off.

(you don’t need to melt all the ice; just the ice next to the surface of the stick in order that the rest can slide off)

I quite like the frozen church cross variant. :slight_smile:

1 Like


Interestingly the Daikins do actually - mine was commissioned by Octopus to uplift the flow temp around 0C (I’ve since disabled it as I felt it made defrosting more frequent due to pushing the unit harder)

Service Manual altherma_3_m.pdf (7.1 MB)


Every day is a school day!

This feature is eminently sensible for mitigating the defrost reduction in heating time. :slight_smile:

1 Like

You shouldn’t really be running a setback when it’s properly cold out IMO.

If the house needs 8kW then run it at 8 kW 24/7 rather than running it at a lesser capacity and then trying to do the reheat part right when it’s cold out and your spare capacity is marginal.

If anything surfing it harder around midday by having a boost period whilst the air is warmest and the grid is least stressed might be sensible?


Nice assessment Colin. One day many years ago caught my old ecodan running in fog at about zero…Supersaturated. defrosting every 25mins. That old model had an incredibly accurate defrost temination. As soon as the last drop of ice melted, it started heating again. Having watched a few newer models defrost, there seems to be a lot of hot air poduced. I’m guessing this is playing safe (every bit of ice needs to go), and in hindsight, it might hve been chance that my old unit had accurate defrost control. Termination for many models seems to be due to one temperature sensor only. I think manufacturers will need to improve defrost termination, mostly because of Colin’s point… if its defrosting a lot, the time needs to be no longer than needed to clear the ice. The ‘blast’ of hot air after defrost is curious. I first thought this a waste… why not delay the fan start a little,… but I guess the ‘bast’ might blow some of the water from the fins.
So, back to the drop in output when +2C. interesing how Daikin show bigger drop than others, and curious that tests at -7 is relatively good. Could it be that at +2 they use 85% rh , but at lower temperatres they use very dry air … dont know.


Good point - storing the heat in the house at midday makes sense, as long as the thermal mass of the house is big enough. Not a ‘night storage heater’ but a day storage heater

Ice - Sticks - reminds me of my old job, in fact they could well be finned tube vaporisers used by the Liquid Oxygen and Nitrogen supply industry. Most users want the gas in their processes so the liquid in the vacuum insulated tanks had to be vaporised first. In sub zero conditions it was a real challenge with larger volume users (over say 5 tonnes/day of LOX or LIN/LN2) boiling temps of -186 or 196c produced a lot of ice. For the big users we used to install two banks and cycle between them. The ice broke off the defrosting bank due to the thermal shock/expansion at the ca 190c temperature rise.
In British Freezing fog oversized evaporators would be a good idea for heat pumps to stop the output collapsing.



1 Like

I think @ColinS intended to type [1] [9] [AltGr]+[Shift]+[0] [C] :wink:

I have mine turned off at night so it has less work to do during times it’s likely to get into defrost. The occupants like it cool when they sleep and the house loses about 3 degrees in 8 hours.

Of course it has to do more work early morning to get going but by then the Sun is shining on it.

Now I’m on Octopus Intelligent with much cheaper power overnight I’ve been running it through the night. The outcome seems to be that it uses more electric and no-one benefits. I’m going to keep doing that for a while to see if it has any positive effects.

1 Like

I agree on this, some setback at night makes sense for my radiator system. 2c temp drop means less heat lost from the house and lower external temps at night would mean poorer cop at night. I like cool bedrooms too and silence. Perhaps it makes less sense for slow responding underfloor systems. When it’s really cold though limiting the setback would avoid very slow heat build up in the morning…