ASHP sizing for Freezing Conditions

@matt-drummer still over-sized I’d say. Averaged over the day the heat generated by the pump was 3.8kw/hr

The house was too warm and had hit 22C so it turned off at 5pm, it then wasn’t needed for the rest of the day. Although it had also started warming up outside by then. Having the heat trickled in to the property would have been much nicer.

The larger Valiant’s and the only Veissman pump seem to have a great modulating range, and/or maybe their systems have been designed better.


Hi Sam,

I managed to get my flow temperature down and maintain a steady temperature in the house.

It was much better in every way.

It ran from 04.00 to 24.00 only stopping for defrosts and the DHW cycle. Over 24 hours we used 80kWh of heat so about 3.4kW an hour.

The temperature is set back to 20c between 00.00 and 04.00 so doesn’t come on at the moment.

It is still a long way under the theoretical minimum output of the heat pump so COP suffers.

It’s also quite a long way from my calculated heat loss, but I think we have all heard enough about that now.

When I start heating my timber constructed office it should all be good, and then I can turn that off I needed to get decent performance in the house (and conservatory!) should the weather get really cold or defrosting becomes an issue.

At least you and us weren’t cold so that’s a big positive.


Similar to my thoughts…an alternative would be a twin system. For me needing 10kw in the house at -2 I would go for an 8kw and a 5kw combo. Turndown without cycling to less than 2kw but still copes with freezing conditions.


Totally agree that we need to avoid switching to direct electric for backup and take into account total system efficiency.

When we talk about over-sizing or correct sizing, there’s one thing we need to keep in mind that any suggested oversizing multiplier is hugely affected by the accuracy of the heat loss calculation or measurement and also how customers choose to use their systems. To illustrate with my own heat pump:

  • When I first did the heat loss calc I came to a heat loss of 5.5 kW, I did not appreciate that air change rates were so low and so used values in the 1.5 ACH range. I also used high U-values for my uninsulated solid stone walls and old double glazing.
  • When I contacted a local installation company to quote they separately suggest an 8.5 kW Ecodan based on the EPC of the house.
  • I had a strong huntch from previous experience that the 5kW ecodan would be more than enough and we have a 5kW wood stove that we could fall back on if there was a problem. (Not that I recommend wood stoves due to air quality issues!)
  • I ended up installing the 5kW heat pump myself for several different reasons: learning, cost, getting the exact design I wanted.
  • The maximum continuous heat over 24h it has delivered is 3.08kW (74 kWh).
  • It’s actually quite hard to accurately measure the heat loss in our house as the thick stone walls have so much thermal intertia that it can take days for outside temperature changes to work it’s way through the fabric.
  • The best I’ve come up with is a monthly calculation that looks at total heat delivered, average indoor temp, average outdoor temp, internal gains from lighting, cooking and appliances, metabolic heat & a solar gains estimate, full details here.
  • On this monthly basis my heat loss parameter varies from as low as 131 W/K to as high as 201 W/K in winter months. It’s typically closer to 180 W/K in winter. This would suggest ~4.1 kW heat loss at 20C inside, -3C outside.
  • Internal gains from electric consumption contributes ~160W, metabolic gains perhaps ~120W, solar gains not that much in winter maybe 20-60W. Reduces heat demand from the heat pump to ~3.8 kW.
  • Outside temperatures are never cold enough for long enough given the thermal inertia in our thick stone walls for the heat pump to really see the heat loss from that -3C outside temperature. We probably only see average temperatures closer to 0-1C.
  • We also choose to run average internal temperatures lower than the heat loss design assumptions with set backs and lower bedroom temperatures.
  • All in all that’s meant we’ve never put in more than 3.08 kW of heat in a 24h period, and usually dont put more than 2.6 kW during near 0C outside temperatures.

So what is the real oversizing factor on my system? and what would it have been if we’d gone with the installation companies initial suggestion of an 8.5kW unit!?

  • Based on 4.1kW measured heat loss without taking into account internal gains, it’s about 1.22x
  • Based on 3.8kW including internal gains it might be 1.32x
  • Actual max delivered over 24h period 3.08 kW = 1.62x
  • Had we installed an 8.5 kW heat pump as suggested = 2.1-2.8x

We need to distinguish between sizing based on an accurate heat loss measurement and sizing based on an inflated heat loss calculation :slight_smile: but all in all getting this right is not always that easy…

Sorry this post ended up being quite a long one! :joy:


The maximum we have observed over 2.65yrs of operation was on 16 December last year. This was the 3rd day of consecutive sub zero temps (average temp -3.6C over 3days) and the heat delivered was 66.91kWh on the 16/12/22, so 2.79kW, although internal temps dipped a little to about 18C (from a more typical 19C for our place).

Big Yes agreed to both of these observations. In our case with somewhat better insulation and also relatively high levels of thermal mass (behind our EWI) then the lag between outside temp and internal heat loss is similar such that we need 2 or 3 days at -2 to -3C to fully observe the “rated heat loss”.

1 Like

One caution about my comments on oversizing to take care of the defrost limitations;

The Three year average temperature distribution in the South east of England (Basingstoke) shows that only 2.1% of days have a 24hr average temp of Zero or less (about 7 days in a year) . 80.2% of the days have temps of 1 to 16c and the rest are Summer days when the heat pump is not space heating at all.

The energy use on the Zero or less days (for me) is 8.3% of the total space heating energy and the 91.7% balance is for the 1 to 16c range. Clearly it is better to optimize the COP for the 91.7% even if you need some additional heat source for the sub zero temps…

In my case it is easy to use one or two small, inexpensive, 2 kw fan heaters in the room or rooms being used. These can easily be switched off when I go out and at night. This is better than using a 3kw back up heater on continuously to boost the heat for the whole house. So despite having only 7.5kw when I sometimes need 9.8 kw it is easy and quite inexpensive to add in the 2 to 4kw needed to heat the house nicely.

The South East has colder Winters than much of South and Central UK as it is closer to the continental ‘hot summers cold winter’ pattern. So if you live in the warmer (in Winter), west of England and Wales (all the way from Cornwall to Cardfiff to say Manchester) then you will have fewer than my 7 days a year sub zero. I would argue that you should use Zero C as your design temp in those areas and not -2c used in the South East.

In summary I’d pull back from my 1.33x nameplate output adjustment to say 1.2X… and as pointed out above that assumes that you have an accurate energy loss calc or like many of us, have looked at your gas consumption records to see how many kw you actually put in the house in past years.


By chance last week I came across the sizing data used for my first Daikin Heat pump installation - 12 years ago. The installer was excellent and really knew what they were doing. It was a 6kw system (with 3kw of backup ) and you can see from the excellent sizing chart below that the defrost issue has been taken into account,

The back up hasn’t been needed but it is a good insurance policy if the heat pump fails.

Why on Earth don’t they do something like this now !!!


Unfortunately those “marginal COP 1” supplementary heaters are only used when the grid is maximally constrained.

Better for the grid would be to fit an AC to the first floor landing (where cold air can fall downstairs if desired; or enter bedrooms; controlled by doors) and in a cold snap shut off the rads upstairs and dial up the flow temp to the rads downstairs. (the heat pump having lower capacity at higher flow temps)

And then time the heat pump to “off” during the peak 3 hours or so per day on those midwinter days that everybody else is plugging in fan heaters.

Or indeed accept 18C internally and wear a hat during sustained cold snaps; whilst scheming how you put a foot of mineral wool on the outside of the walls. Or not if the climate continues to change.

Agree that southeast is closer to continental weather.


I have one of those. It’s called “Termowentylator” in our house.

I used it in my office before we had the heat pump. I often work during the night in my office so didn’t want to waste energy heating the whole house.

Then I experimented with the heat pump. As you can see from last night, the heat pump uses 1.2kW to keep the whole house ticking along:

So now I’ve decided to use the heat pump during the night instead of termowentylator.

As we all know, mine is oversized for most of the time so I do get a lower CoP but I mitigate that by switching it on and off. I have the luxury that the Ecodan gets up to reasonable efficiency relatively quickly.

Just for the record, it did actually manage a long run yesterday and it was pretty underwhelming efficiency. We were warm though. The effective temp was just hovering below zero.

You might argue our heat loss is about 4.6kW@0 °C from that chart so having a 14kW heat pump is clearly much bigger than required.


Don’t you need planning permission for a second heat pump though??

……and there is the irony, and lack of joined up thinking (system design!) in the UK

…and yet a fossil gas hybrid system is somehow allowed and also acceptable!


You need planning for ANY air to air heat pump.

You also need planning for ANY water to air heat pump…

…unless it is being installed by MCS union labour.

That’s the rules that they have been permitted to write for themselves and that’s Ian Rippin’s stated objective. (ensure that there is only one way that heat pumps can be installed and that MCS governs that way of installing heat pumps)


Interesting discussion, I’ve just had my heatpump installed, An Ecoforest EcoAir 3-12, but aside to that.

I am wondering when it does get to minus 2 (or below), the humidity should be much lower, and then the defrost cycles do less harm to the performance.

The problem is that at -2c the house heating requirement will be higher than at zero c so the heat pump will be working harder, with a larger DT between evaporator and outside air. If yours is oversized as many seem to be, then it still shouldn’t be a problem.

My system is at it’s limit at 0c - and reaches what I think is the maximum defrost frequency 30 mins on and 6mins defrosting, defrost frequency doesn’t seem to get worse at -2c but it does not produce any more power for me either.

I suspect you are right if you have a very low humidity sub zero day but that is not usual in our maritime climate.

1 Like

I’ve posted another test result from my EcoDan from last night, it met the datasheet output of an average output of 5kW and COP of 3, though time between defrosts seemed quite long! not really conclusive enough yet, just another datapoint: 5kW R410a Ecodan PUHZ-W50VHA2(-BS) maximum output testing - #7 by TrystanLea

@glyn.hudson also did a test at the same time on his Samsung 5kW (in the same village so similar conditions) and also achieved the same match to datasheet capacity 4.5kW on his, full details in his thread here



Another test result this evening, mine ran ~7.5% below datasheet suggested capacity over 3 hours at 0.84C outside 4.57 kW rather than 4.94 kW, I’m being overly precise here… roughly 10% below datasheet.

@ColinS had less favourable results 20-30% below I believe.


Whilst we’re sharing :slight_smile:

Here’s a rare sight of my oversized Ecodan 14kW actually running all the time (can you see an Octopus Saving Session?)

Nominal CoP@Flow is around 270% and it did 285%. In the “Efficiency” chart you can see the gold line showing the manufacturer’s nominal CoP@flow:

When it’s trucking along it gets about 300%, but there’s a lot of defrosts in there so it’s paying the price of getting less efficient, then heating the ice and then getting going again.

You can see things went wrong around 14:00 when the air was getting warmer and it didn’t know whether to give up or not.

The outside temp was between 0 and 2 °C but the effective temp (which my control algorithm uses) was mostly between -3 and -1 °C. Dew point was -2 to 0 °C.


Here’s @ColinS’s test from yesterday evening:

  • Outside temp back of the unit: 1.32C
  • Metoffice: 1.9C
  • Metoffice dewpoint: -0.9C
  • Metoffice relative humidity: 82%
  • Average output: 8.7 kW
  • 1st cycle: 9.9 kW (2.3 C outside)
  • 2nd cycle: 8.7 kW (1.4 C outside)
  • 3rd cycle: 8.2 kW (0.6 C outside)
  • 4th cycle: 7.8 kW (0.6 C outside)
  • Flow temperature peaked at 42C in the 4th cycle but averaged closer to 40C.
  • Target flow temp was set to 50C to give the heat pump a clear signal to push as hard as possible.
  • No backup heater running

Databook suggests output should be between 10 kW and 11 kW at these temperatures perhaps around 10.5 kW? So we are seeing 22-26% lower continuous outputs in those 0.6C cycles it seems:


Is that fair @ColinS ?


Spot on Trystan. It’s worth pointing out that my radiator capacity (including 4 convectors) is 12 to 13 kw at a mean flow/return of 40c so the system is not emitter constrained. My house was definitely getting warmer last night with around 8.5kw input at about 1c so it is close to my ‘design’ capacity requirement of 9.8kw at -2c… which the Daikin data book says it should deliver. Grrr!

This evening with the same outside temperatures I am pushing it less hard - just 6.8kw and the COP has recovered from 2.3 to 2.9 along with larger intervals between defrosts.

I am still puzzling over the tech data Daikin supply - two pages after the capacity graph in the post above comes this table with wildly different capacity values and COPs for the same machine.

I took the MCS number and the graph as accurate when dealing with Daikin but now wonder what these other test standard results really mean. Do the other brands produce a similar tables to this?


I noticed that a “show defrosts/ cooling” button has appeared, is this recent?
damage on my system for sat 2nd when it defrosted every 45 mins, -2C to -0.5C.
instantanous COP during the (short) stable periods is about 2.7, overall is 2.3 so the 10% hit on energy hits around 15% on COP.

1 Like