Heat Pump Energy output over a 24hour period as a function of different variables

Dear all,

I’ve been trying to get my head around how the maximum energy output in a 24hour period changes for a given heat pump in response to the different variables, most critically the defrost cycles, which seem to radically reduce output (5kW rated heat pump down at an average of 3kW on a cold, wet day, see attached image)
This is specifically with regards to selecting an appropriate size of heat pump, I understand the internal emitters need to be able to sink the energy output but I’m interested in the source side for now as I understand that less well.

It’d be great if someone could confirm/ correct my understanding of the various relationships as below:

  1. Energy output falls as outside temperature falls and/or as flow temperature rises. (the heat pump COP reduces when pumping across a higher temperature difference and the compressor has a max power in)
  2. Energy output falls when the outdoor humidity rises and or temperature falls due to defrost cycles kicking in that require the heat pump to instead extract heat from the house (or at least not pump heat into the house for a period of time) (It seems to me from looking at other peoples data, a 5kW rated heat pump could be down at 3kW average output from this effect during cold, wet days)
  3. Energy output falls whenever the heatpump is switching on and off for any other reason.

So I understand the principle behind trying to avoid oversizing a heat pump…you want to avoid short cycling where it can’t modulate down sufficiently (lots of energy loss from cycling on and off), and you want to be running it at its rated output because everything is designed to be most efficient at that point.
An interesting side point is that it seems that bigger output heat pumps have bigger fans and a bigger heat exchanger so I’m assuming that a bigger heat pump will be quieter on average for a given heat demand than a smaller heat pump (likely will consume more energy due to short cycling but will be quieter?)
I imagine that although a larger fan is more efficient, the fan uses such a small amount of the total energy (the vast majority is used by the compressor?) that the incremental benefit of more efficient air flow over the heat exchanger doesn’t offset the reduced efficiency from short cycling in the slightly warmer months (March/ April) when a smaller heat pump would still be able to modulate down enough to run continuously?

I’m also imagining that newer heat pumps can modulate down much more effectively and so oversizing isn’t such a problem?

Thanks in advance for anyone who helps with clarity on this!

1 Like

Hello @Henry_Fletcher

This is a really important question I think! I thought I’d try and crunch the numbers on the heatpumpmonitor.org dataset. Not all of these have complete datasets for the winter, I’ve removed systems that have only joined after the latest cold spell (~8th March 2023).

  • kW1 is the maximum recorded output over a 24h period (total kWh heat delivered divided by 24)
  • %1 is this value as a % of heat pump capacity.
  • OT outside temperature

I’ve also included the second and third highest output days.

The heat loss value as given by the survey and head demand / 2.9 rule figure is also shown for context:

I think it really highlights the difference in how manufacturers rate their systems?!

location heatpump output heatpump model heat loss survey heat demand div 2.9 kW1 %1 OT kW2 %2 OT kW3 %3 OT start date
Oxford 3.5 Vaillant Arotherm+ 3 7500 2.6 3.79 108 -1.5 3.79 108 -2.4 3.75 107 0.6 02 Dec 2022
Heat Geek Towers, Camberley 5 Vaillant Arotherm+ 5 6000 2.1 4.96 99 -2.3 4.72 94 -0.4 4.51 90 -1.2 19 Nov 2021
North Hertfordshire 5 Vaillant Arotherm+ 6 8000 2.8 4.38 88 -5.2 4.08 82 -3.1 4 80 -4.6 16 Nov 2022
Stratford-upon-Avon 7 Vaillant Arotherm+ 7.57 7300 2.5 5.92 85 2.1 5.63 80 1.2 5.46 78 0.4 01 Feb 2023
Banbury 7 Vaillant Arotherm+ 6.5 0 0 5.92 85 2.1 5.63 80 1.2 5.46 78 0.4 01 Feb 2023
Sheffield 5 Vaillant Arotherm+ 5.4 9000 3.1 4.15 83 -0.8 4.1 82 -0.6 4.03 81 -0.2 12 Oct 2022
Newtown, Wales 5 Panasonic ‘J’ Aquarea 11 26000 9 4.07 81 4.07 81 3.79 76 08 Nov 2022
Telford 5 Vaillant Arotherm+ 4.5 11236 3.9 3.88 78 3.5 70 3.29 66 23 Dec 2022
Llanberis, Gwynedd 5 Samsung Gen 6 3.5 8000 2.8 3.58 72 2.8 3.16 63 5.4 3.12 62 2.1 10 Aug 2022
Fife, Scotland 7 Vaillant Arotherm+ 0 23500 8.1 4.87 70 -4.7 4.81 69 -4.4 4.74 68 -2.4 09 Sep 2021
West Yorkshire Peak District 8.5 Mitsubishi Ecodan 8 15452 5.3 5.74 68 -2.2 5.08 60 -0.7 5.06 60 -1.3 20 Aug 2017
Cambridge 8 Samsung HTQ 7.8 10897 3.8 5.08 64 0.3 4.42 55 4.5 4.29 54 4.5 31 Jan 2023
Basingstoke 10.6 Daikin 11kw monobloc 9.8 21000 7.2 6.69 63 0.9 6.65 63 5.9 6.11 58 0.8 13 Dec 2022
North Yorkshire 6 Grant Aerona3 6.25 15500 5.3 3.71 62 2.6 3.21 53 -1.2 3.2 53 6.8 21 Oct 2022
Manchester 5 Mitsubishi Ecodan 3.4 9500 3.3 2.73 55 8.3 2.64 53 0.8 2.56 51 0.9 19 Sep 2022
Gloucestershire 5 Mitsubishi Ecodan 3.5 7200 2.5 2.75 55 3 2.67 53 2.9 2.61 52 -1.1 20 Jan 2023
Bristol 7 Vaillant Arotherm+ 8.3 15800 5.4 3.85 55 3.6 51 3.56 51 27 Jan 2023
Weymouth 14 Mitsubishi Zubadan 0 27328 9.4 7.63 54 3.3 6.63 47 3 6.5 46 3.3 18 Nov 2022
Nottingham 12 Midea 8 11332 3.9 6.41 53 0 6.07 51 -0.9 5.87 49 -0.7 08 Jan 2023
Banbury 8.2 Vaillant Arotherm+ 7.6 0 0 4.33 53 3.1 4.3 52 3.5 4.2 51 4.1 17 Feb 2023
Llanberis, Gwynedd 5 Mitsubishi Ecodan 4.5 8000 2.8 2.64 53 4.5 2.62 52 -1.4 2.61 52 -1.2 28 Sep 2020
Weston-super-Mare 7 Daikin EDLQ07CV3 0 9524 3.3 3.57 51 -0.4 3.55 51 -1.7 3.3 47 1.5 03 Oct 2022
Swansea 11.2 Mitsubishi Ecodan 0 30000 10.3 5.51 49 4.3 4.33 39 4.8 3.93 35 5 25 Feb 2023
Manchester 6 NIBE F2040-6 3.5 7000 2.4 2.8 47 -1.9 2.8 47 -3.6 2.71 45 0.2 03 May 2022
Bangor, Wales 11 Mitsubishi Ecodan 5.5 11000 3.8 5.21 47 2.7 5.16 47 -1.4 5.09 46 -2.8 22 Sep 2022
Malton, North Yorkshire 14 Mitsubishi Ecodan 14 41699 14.4 6.47 46 -1 5.7 41 2.4 5.6 40 2.6 15 Dec 2022
Sofia, Bulgaria 8 Atlantic Alfea Extensa 5 12800 4.4 3.67 46 -5.8 3.63 45 -7.2 3.58 45 16 Dec 2022
North Yorkshire 11 Mitsubishi Ecodan 9.4 22000 7.6 4.72 43 -0.4 4.64 42 -0.4 4.62 42 -1.5 11 Jan 2022
Sheffield 11.2 Mitsubishi Ecodan 9.989 23594 8.1 4.68 42 1.3 4.54 41 1.9 4.2 37 0.9 05 Mar 2022
Kidderminster 7 Acond PRO-N 37 6570 2.3 2.96 42 3 2.71 39 10 2.54 36 5.1 21 Feb 2023
St Albans 5 Vaillant Arotherm+ 4.1 9143 3.2 1.96 39 5.3 1.79 36 6.2 1.75 35 6.9 08 Mar 2023
Machynlleth, Wales 6 Mitsubishi Ecodan 1.5 7500 2.6 2.23 37 11.7 2.04 34 -4.5 1.96 33 15 10 Feb 2022
Newbury 14 Mitsubishi Ecodan 11.5 12000 4.1 5.24 37 -2.4 5.04 36 -1.1 4.92 35 -2.2 13 Nov 2022
Derby 8 NIBE F1145 0 4542 1.6 2.67 33 0.7 2.54 32 1.3 2.46 31 3.4 11 Dec 2022

Download as ODS:
heatpump_output_vs_capacity_v2.ods (19.4 KB)


Yes, I think this is true. Vaillant certainly underrate their units while other manufacturers overrate their units. There should really be a standard for nominal output. It’s even more confusing sice some heatpumps share components e.g a 3.5kW Vaillant uses the same compressor as a 5kW unit, while a 8kW Samsung HTQ uses the same compressor as a 14kW unit therefore the capacity of these units are much higher but the high-end modulation suffers.


This is probably true, but hard to quantify since most manufacturers don’t publish information about the lowest modulation.

Based on my monitoring I can see that my 5kW Samsung Gen6 will modulate down to about 3kW, in the summer the lowest modulation will be higher than in the winter. However, cycling does not seem to reduce efficiency that much, I actually found that running my heatpump at a slightly higher flow temperature 35C instead of 30C resulted in a slightly higher overall COP, this resulted in more cycling but a higher Carnot efficiency.

e.g here’s a typical day in the shoulder heating month of April. The default room temperature control on my Samsung is quite loose i.e it will allow + or - by 1C above and below the set point, a tighter temperature control would result in more cycles of shorter duration.

Zooming in on one of the cycles, very little energy is wasted at startup since the HP starts producing heat output almost immediately after startup.

If you don’t have backup heating source e.g wood stove, a modest oversizing e.g 20-30% is probably a good thing to ensure you can meet full demand including DHW in winter.


Thinking through the data above

  • we know heat loss calculations have large error bands and so the building heat loss figure is probably at least ±1 kW if not more.
  • The rated heatpump capacity listed is just that, it’s rating, not the datasheet figure for the given outside temperature.
  • Outside temperatures for these max 24h periods vary considerably from -5.8C up to 11.7C (the higher end could probably be filtered out for this analysis as we are mostly interested in cold weather performance).
  • We don’t know average internal temperatures and so cant say what % of design heat loss the house was experiencing.

Those caveats aside:

  • There are some impressive figures in there from the Vaillants, getting a very high % of their rated capacities. First three in particular at below zero temperatures.
  • A lot of systems (mine included) in the 40-60% of rated capacity range
  • A lot of systems putting in less heat over 24h than design heat loss - but we dont know internal temperature and outside temperatures are often higher than -3C… and so cant say what % of design heat loss the house was experiencing.

Well done Trystan… It will take me a while to get my head around it all… I know that I had looked for ‘worst defrost’ period for my old 5kW Ecodan. I found a period of freezing fog where it defrosted every 40mins. the output during this period was well below 5kW. This of course skews the figures. For the R32 6kW, it has the ability (and annoying habit) or revving up to over 8kW, so I guess had no problem achieving its rating in the rare occurrence of freezing fog.


Dear Trystan Glyn and John,

All very interesting and useful information, thanks for crunching that data Trystan.

I guess I’m perhaps most interested in the worst case performance (i.e. lowest energy output over 24hours, rather than the highest) when it was a cold, damp day AND (critically) the heat pump had a constant demand signal (or alternatively if it only had a heat demand for 12 hours then divide the energy output by 12 rather than 24).

What is the easiest way to download the dataset for analysis Trystan? Is this something anyone has permission to do?

@glyn.hudson, I seem to be in a slight battle with installers who want to install 12kW systems in our house despite the fact we have two wood burners and even when we aren’t using the wood burners we rarely used more than a total of 60kW of energy (gas and electricity) for everything (water heating, space heating, electrical appliances).
I often get the response that if the survey says we need a 12kW system then we have to have that installed or we can’t benefit from the governments £5k boiler upgrade grant scheme.

I appreciate that we have our house on the cool side and a future owner might want it hotter (and not want to replace the heat pump) but it irks me slightly that things are this rigid with the grant, especially as we plan to live in this house for a lot longer than the heat pump is likely to last!

I must say that the Vaillant Arotherm + range is looking like a good bet right now, I especially like that it is R290 rather than R32.

Heh heh.


It’s an obvious and leading question but why not forgo the BUS grant and just buy the system you actually want?

The BUS grant is to assist with environmentally friendly heating solutions. I can understand that the government take the view that supplementing a heating system that they have contributed quite heavily to with two wood burning stoves doesn’t fit the remit of the BUS grant scheme.

Obviously, the aim is to stop burning stuff in our homes.

If you have a wood stove it’s obviously a much lower risk to sail closer to the wind with regard to sizing. The installers I’ve spoken to have said that MCS is more flexible with sizing than many installers will let you believe. There are so many assumptions during the heatloss survey, it would probably be quite easy for an installer to adjust some assumptions e.g reduce air changer per hour to reduce the heat loss figure and therefore install a smaller but correct on paper HP. But I don’t have experience of this myself since I self installed.

1 Like

Trying to think how we could extract that? Outside humidity levels is something we are missing in the monitoring so far, perhaps something to explore recording via MetOffice api in future. We dont specifically record the demand signal. But can extract stats for the period above a certain consumption threshold…

It might be interesting to try and find specific periods from the public dashboards manually and work backwards to think how such an analysis could be automated?

The dataset is split out over individual dashboards and accounts, it’s not that easy to access at the moment apart from via the public dashboard links, we will be doing more around improving the API for this and documentation in time… there probably close to 4 GB of data so far behind those dashboards so finding efficient ways of making relevant data available is something that needs work as otherwise server load from downloading that data could be an issue.

We probably need to work out a way of comparing the monitored data more directly with manufacturer datasheets, to see if we generally replicate what the datasheets suggest or find discrepancies to do with real world vs test procedure conditions. This will likely vary from manufacturer to manufacturer and it is often difficult to untangle the heat pump units data from the wider balance of system performance… so not an easy thing to do

1 Like

@matt-drummer, yep, am considering purchasing it outright/ possibly doing the install myself. With regards to wood burning, I should do some calcs, I imagine everyone burning a few logs on a wood burner on the coldest days every winter might be ok sustainability wise (depending how the timber is sourced). Maybe, maybe not. The comforting flicker of a fire is a hard to part from luxury on a cold winters day! Certainly burning a lorry load of coal at everyone’s house every winter is not sustainable in any way!

@TrystanLea. Probably can get a reasonable proxy for humidity retrospectively?(Weather in January 2022 in London, England, United Kingdom)

Looking at the graph from the public dashboards that I shared in my original post, one can fairly easily see where the demand starts and ends, as you say, a threshold on the modulus of the heatpump power out, maybe with some hysteresis would probably work well to approximate the on demand in a way that could be automated across the dataset.

I guess my basic goal is to understand the pros and cons of oversizing more intimately. Households like ours are increasingly going to have accurate consumption data of their energy usage via smart metering. I can see how much gas and electricity we used on average across the winter and also on the peak days…I’m interested in what the reasons could be for installing a heat pump that is well over capacity for even the peak days?

  1. As heat pumps get older their output falls? Does anyone have any data on this?
  2. When it is cold and wet, heat pump rated output cannot be sustained over a 24hour period due to defrost cycles etc? How much is this reduction? If I can run my 5kW rated heatpump at a 4.2kW output on average for 24hours I can get 100kWh. I think houses need to be pretty big/ poorly insulated before they require more than 100kWh in one 24hour period, so I’m not sure why anyone is fitting anything other than 5kW and smaller rated units!
  3. Small heat pumps running hard are noisy compared to large heat pumps running at a low modulation?
  4. If I like my house at 18 degrees C and I size a heat pump for that, but a future owner likes it at 23 degrees C…they might not be able to get the temperatures they want with the heat pump I spec…I guess that’s up to the homeowner to accommodate it or not if they are paying for the whole thing, but I can see why the government might not want to give grants for lots of heatpumps that then get pulled out.
  5. Fear from installers that customers will complain their house is too cold if they install too small a heat pump…

I have a feeling that number 5 is the predominant reason in most cases that 11kW rated units are getting installed when a 5kW unit would be absolutely fine…but I’d be interested in your thoughts?
Maybe my figure of 100kWh in a 24 hour period is really small and actually many many houses use more than this? Though a bit of googling would suggest that 100kWh is likely above average based on: https://utilityswitchboard.com/gas-electricity/guides/tariffs/average-usage/
12000kWh of gas, let’s say 10,000 get’s used across 5 winter months, that’s about 66kWh a day on average, not accounting for gas boiler inefficiencies, but also that’s an average not the coldest day.

It’s not just defrost cycles that take away from heating as you also need to heat hot water.

I would think that the sooner you can get the hot water heated each day, if that’s what you do, the sooner you can get back to heating.

I used around 13,000 kWh of gas last year.

My worst day used 110 kWh of gas and that includes heating the hot water.

Assuming the best case of 90% efficiency from my gas boiler I need about 100 kWh on the worst day.

Our house is around 120m2 including a conservatory (that is not as difficult to heat as you might imagine).

Octopus have installed a 9kW Daikin Altherma 3 here.

It is bigger than I thought we needed as I thought a 7kW Vaillant would be about right.

Our heat loss was around 6kWh at 21c inside and -2 outside if I remember correctly.

I think what we have will work well and we won’t be cold.

I a looking forward to finding out what it will cost to run but I am not too concerned.

I have read somewhere that 1.2 to 1.5 times the heat loss is not a bad solution.


I was having so much difficulty getting someone to install here that I considered self installing. I could have done it, I can solder and I can do electrics if I needed to.

The reality is that I probably couldn’t have even bought the heat pump for what I paid Octopus.

The amount of stuff required is astonishing and all the little bits add up quite quickly.

Then there’s all the other stuff, their support is amazing, I pay £9 a month for them to look after it, it’s guaranteed for seven years and they did a fantastic job.

Apart from not being offered the openenergy monitoring I couldn’t be happier.

I don’t think anybody could have or would have done a better job however much I had paid.


Ah, a very glowing review of the Octopus installers indeed! Do you mind sharing how much it cost you for them to do the install out of interest? Did they just install the heat pump or add a hot water tank and new radiators too?


I am happy to answer any questions but maybe it’s not entirely relevant to your thread.

I could start a topic on it if anybody would be interested, I am very proud of what they did!

I only mentioned it because you were talking about a self install.

I had two radiators changed, they are quite large so I would have thought they added £700 or so to the costs, I would have thought the two radiators cost at least £700 if I wanted to buy them let alone fitting them.

I paid £6,700 net of the grant.

I got the Daikin Altherma 3 EDLA09D3V3 heat pump
A Joule 250 litre tank
World Heat 20 litre volumiser
Dedicated consumer unit for the heat pump
Upgrades to my meter cabinet
Removal of my gas boiler and old vented hot water system
Upgrades to the radiator plumbing where necessary
Removal of my gas meter
They put the external pipework in rather expensive trunking at my request in addition to the original quote, they charged me precisely nothing for it!
Various trips to suppliers to buy bits and pieces just to make it better, things like a grill for my old gas boiler flue hole rather than just cementing it up, lots of little touches that I don’t think any other installer would have done.

An installer came back after three weeks to check the quality of the installation and clean the magnetic filter and the filters in the heat pump. There was some `gunge’ from the solder flux and stuff like that.

I have never heard of anybody talking about service like this.

I have a seven year parts and labour warranty and I pay £9 a month for the service plan.

I apologise for gushing about it, but they were so good, not a single complaint, not a single tut, they were a joy to have in the house for a week.


That’s really interesting to know, thanks for sharing @matt-drummer, maybe I’ll ask them to quote for ours too! Out of interest, was it an “Octopus” team (Octopus labelled van, Octopus logo on their shirts etc) or did they contract to the best installers in your area I wonder?

I would say Im edging towards the view that slight over-sizing might not be too bad. Especially if there’s no backup heat source like a wood stove… or perhaps right-sizing on a vaillant would be fine vs a need to be more careful and ensure enough head room on say a daikin…?

We’ve seen a few examples such as Glyn’s suggesting short cycling may have relatively little performance impact… though it’s not clear if the accuracy of that result is impaired by the relatively long 10s measurement interval of the monitoring vs the speed at which the underlying conditions are changing. It would be interesting to validate a short period at say 1s resolution but then you need temperature sensors that can respond much more quickly.

The key of course for high performance is to design for low flow temperatures and then to get the controls setup correctly.

I’m wondering of the poorly performing heat pumps out there, how many could be improved with some relatively simple and free control settings adjustments. We managed to improve the performance of one local 11kw ecodan (5kw head demand) by +1 cop with simple control changes, including removing the individual room zoning… house is much warmer too. how representative that might be of wider installs Im not sure.


@matt-drummer that’s great to hear such a positive experience!!

1 Like


I’m exceptionally interested in the effects of room zoning vs COP vs total energy consumption for a given comfort level for occupants.

I am aware of situations where people heat their whole (large) houses with a heat pump, get a good COP but use a pretty huge amount of energy per person.

Certainly with our gas boiler, heating our whole house chomps through gas like nobody’s business, if we just heat the rooms we are in, we use something like 25% of the energy vs heating the whole house…I commented on a heat geek video about the same thing, I think this is very dependent on usage patterns and the actual layout and structural materials/ insulation of the house.
I can definitely imagine situations where the increased COP from heating more of the house (especially with a larger, older (worse at modulating) heat pump) could more than compensate for the extra output energy required, and so use less input energy in total…but I also think that in many situations, more input energy is actually required…sure the COP goes up…but I’m not sure that optimising for the best COP should be the primary objective…minimising energy input per person (specifically high CO2 intensity energy) for a given comfort level seems to be a better metric in my mind.

1 Like