Friends, I have written a lot on this blog on the subject how to choose the “right” size of heat pump for a dwelling. Normally this means installing a heat pump which can meet 100% of the heating demand on the coldest day. For the South of England, that so-called “design-day temperature” is typically -2 °C or -3 °C. Typically one would like to have a little margin, so a heat pump which can supply (say) 120% of the maximum anticipated heating demand would probably be considered “about right”.
When I was having a heat pump installed I was worried that the device would not be able to heat my home and that my partner would be unhappy. Similar concerns of clients about partners, and installers about clients, mean that over-sizing is commonplace. No one wants to be cold!
But this article is about installing heat pumps which will definitely not be able to heat a dwelling on the coldest anticipated day. Strangely, in the right circumstances, this can be a smart option.
Click on image for a larger version. Based on 3 years of daily data from Heathrow Airport, the graph shows the typical number of days per year that the given fraction of full heating power is required. Full power is the power required to keep a dwelling at 20 °C on the coldest day in the last three years. Adding up the individual data points, one can see, for example, that for typically 69 days per year, the required heating power is between 40% and 60% of the maximum. Heating power between 80% and 100% is only required on 12 days per year.
To see why, imagine a “perfectly-sized” heat pump that will meet 100% of heating demand on a day at “design temperature”. The graph above is from this article and shows that the peak heat pump power is only required for perhaps 1 or 2 days per year. If the heat pump is 20% oversized, then the peak power will likely never be required.
Indeed a heat pump sized to supply only 80% of the design-day heating requirements will heat the house adequately for all but (on average) 12 days per year. This means the dwelling may need an additional kilowatt or two of supplementary heating power, for a week or two per year.
So in cases where installing a heat pump with additional power causes some additional complexity or cost, it can make sense to deliberately under-size a heat pump, and just add some electrical heating to cope with the coldest days.
In this article I will firstly estimate the likely impact of under-sizing in terms of reduced SCOP and increased running costs, and then point out some of the circumstances where under-sizing makes sense. And finally I will tell you a story: a really interesting true story about over-sizing and under-sizing and happy customers.
The Impact of Under-sizing
In order to estimate the impact of under-sizing I have created a spreadsheet model (download link) that compares two heat pumps heating two identical properties. The two heat pumps are assumed to be identical, except in the heating power available. The assumptions are summarised in the graph below:
- The Fully-Sized heat pump is assumed to have a heating power of 10 kW at 20 °C external temperature, and 8 kW at -5 °C.
- The Under-Sized heat pump is assumed to have a heating power of 7 kW at 20 °C external temperature, and 5 kW at -5 °C.
- Both heat pumps share the same COP. For space heating the COP is assumed to fall from 5.0 to 2.5 across the range, and for hot water it is assumed to fall from 3.0 to 1.5 across the range.
Click on image for a larger version. Graphical illustration of the way the maximum output power changes with outside temperature for the fully-sized heat pump and the under-sized heat pump. The COP also varies with external temperature but this is assumed to be the same for both heat pumps.
The spreadsheet assumes a particular value of Heat Transfer Coefficient (300 W/°C) and uses this to calculate the heating required with external temperatures ranging from -5 °C to 20 °C. In the circumstance where the under-sized heat pump cannot provide enough heat, the balance is assumed to come from direct electrical heating. This could be either done informally (turning on a fan-heater when people feel cold) or be built in to the heating system so the occupants will not notice any difference between the two systems.
Click on image for a larger version. The graph shows how the coefficient of performance (COP) varies with the amount of heat pumped per day. For the under-sized heat pump, the COP falls at high demand because the heat pump is running at maximum power and being supplemented with resistive heating with a COP of unity.
The figure above shows how the COP varies with the amount of heat pumped. The systems perform identically until the external temperature falls below roughly 0 °C. Then the COP of the under-sized pump falls as the resistive electrical heating supplements the heating from the heat-pump.
The spreadsheet also considers how often each external temperature would likely occur throughout the year based on data for the climate around Heathrow airport. The heat delivered to the home (20,236 kWh/year) is the same in both cases.
Assuming that a gas boiler were used to provide this heat with 85% efficiency, this would require 23,807 kWh of gas annually, and the Heat Pump Rule of Thumb would suggest a heat pump with a heating power of 8.2 kW would be about right.
Combining the heat pump model with the estimates of the number of days at each temperature allows for estimates of the annual electricity consumption, cost and Seasonal Coefficient of Performance (SCOP)
The under-sizing causes the SCOP to fall from 3.36 to 3.14 and the annual cost increases by around £100 from £1,471 to £1,574 at an assumed electricity price was 24.5 p/kWh. If the heating had been done solely by direct electric heating the cost would have been £4,957 and if the heating had been done by gas the cost would have been £1,476, just £4 more than the “correctly-sized” heat pump.
Click on image for a larger version. This graph shows the same as the graph at the head of the page but additionally shows the distribution of heating demand in this model as a dotted red line.
Where could under-sizing make sense?
I chose the example above because it is typical of many UK homes. However, many variables can push the balance of costs one way or the other. For example, I have considered the climate around London, but there are many places which have periods of extreme winter cold, and building a heat pump to cope with these extremes can make the heat pump expensively over-sized for most of the heating season.
Whether that makes sense or not is really up to the installer and their client. Adding back-up electrical heating might allow the required reassurance to install a marginally-sized heat pump, and the back-up might never be used. But considering this as an option might help to reverse the widespread tendency towards over-sizing in the UK.
This is actually a specific instance of a classic engineering problem: how to balance the benefits of a more complex or costly installation with lower running costs, against the benefits of a simpler installation with higher running costs.
I asked installer Christoph Grossbaier from Econic about this and he suggested two cases where he considers electrical back-up as an option. When heating demand is close to needing cascading heat pumps, and when noise restrictions affect what can be installed. Both these circumstances affect larger installations, but there may well be other circumstances that might make sense for smaller installations too.
A story about oversizing and undersizing
Back at the start of 2024 I was contacted through friends of friends for advice about installing a heat pump in a large victorian house in Twickenham. I recommended that they get a heat loss survey, but at the time it was cold and I suggested they read the gas meter each day to get a sense of how much heating they were using (Link 1, Link 2). Fortunately, one partner was a data scientist and their astute analysis showed that when it was – 5 °C outside, their home required about 8 kW of heating.
Eventually the heat loss survey was done, but incomprehensibly, despite being told of the heating measurements, they assessed the overall heat loss at 18 kW and recommended two 12 kW pumps i.e. 24 kW of heating power. The survey was clearly £400 worth of nonsense. Eventually, a competent survey was done by Econic which more-or-less agreed with the experimental data and it was decided that after some extra insulation in the floor and one exterior wall, a 10 kW Vaillant Arothem plus would be appropriate.
Three sizes of Vaillant Arotherm Plus Heat Pumps
At this point the clients realised that a 10 kW pump would have a volume of 0.8 m^3 which is more than the 0.7 m^3 volume which is allowed without a planning application. As I mentioned previously, it took two months to get a reply from the council which then waved through the application. But by that time, they had reflected that a smaller pump would be less intrusive in the space available:
- 10 kW pump: 1.6 metres tall: sound level 58-60 dB
- 7 kW pump: 1.0 metres tall: sound level 53-55 dB
And thus a 7 kW pump was installed with an additional 3 kW back-up heater. And the installation went from a clear factor 3 oversizing, to about 20% undersizing with back-up. But hopefully the clients will be warm and happy – and better off than they would have been.
Summary
Appropriate sizing is an important issue for heat pump installations. But concern about sizing has led to widespread over-sizing of heat pumps. This in turn has led to issues with heat pumps under-performing due to excessive cycling during the large fraction of the year when their high power is not needed.
I wanted to write about this issue because, with an electrical back-up, under-sizing (or marginally-sizing) a heat pump installation need not be a disaster. And indeed, in some circumstance it could avoid a good deal of expense and hassle.
Protons for Breakfast 
October 9, 2024 at 12:34 am |
I agree with you. Oversizing a heat pump is almost as bad a mistake as undersizing it.
Bigger is not always better. Oversizing a heat pump can make your power bills higher and/or your house less comfortable, due to the hysteresis problem.
Rapid cycling (turning heat pumps on and off frequently) is inefficient, and it excessively wears the system. So thermostats have built-in hysteresis (a/k/a “dead zone” or “dead band”) to prevent that.
If you’ve set a target temperature on your thermostat, say 23°C, the heat pump will not shut off when indoor temperature reaches the target. Instead it will keep running, typically to about 0.5 to 0.9°C beyond the target temperature, before shutting off.
Likewise, if the heat pump is off, when the temperature passes the target temperature in the other direction, the heat pump will not start up immediately. Instead, it will wait until the indoor temperature is 0.5 to 0.9°C from the target, before starting.
If there’s 0.5 to 0.9°C overshoot in both directions, that means there’s 1.0 to 1.8°C of hysteresis. In winter and summer your indoor temperatures go up and down by about that much all day, and you probably don’t even notice it. (That much temperature change is even less noticeable outdoors, which is one of the many proofs that global warming is not a problem.)
A problem with oversizing a heat pump is that an oversized heat pump will be able to heat or cool your home by 1.0 to 1.8°C <b><i>too quickly.</i></b> So either it will have to cycle on and off very often, at a cost of low efficiency and and excess wear on the system, or else the thermostat will have to increase the hysteresis, perhaps to the point where indoor temperature fluctuations are noticeable, thus sacrificing your comfort.
Here in North Carolina (and throughout the American South), heat pumps have been commonplace for half a century. The primary advantage of using heat pumps is that we all need air conditioning anyhow, due to our hot summers, and a heat pump is little more than an air conditioner with a reversing valve. So a heat pump is the cheapest type of system to install in a new house, here. In fact, in many neighborhoods they don’t even bother to run gas lines.
Here in North Carolina, our heat pumps run almost as hard in summer as they do in winter. (But right now, in early Autumn, my heat pump hardly runs at all.)
Our heat pumps have (very expensive to run!) auxiliary electric resistive heating elements built in, to warm us on frigid winter nights. That would still be necessary even for a severely oversized heat pump, because nighttime winter temperatures often drop well below freezing, and heat pumps simply cannot maintain such a large difference between indoor and outdoor temperatures.
Some people have a combination of AC for summer cooling and gas heat (or occasionally oil) for winter heating, but I don’t think I’ve ever seen a heat pump with gas for the backup heating. That would probably be a very economical system to run, because gas is much cheaper here than it is in the UK (thanks to fracking!), but such a system would presumably be more expensive to install.
For even greater economy, if your local geology is favorable, a geothermal / ground source heat pump is the most efficient and and economical-to-run system of all. Unfortunately, they’re a lot more expensive to install and repair. I understand that former President George W. Bush famously has one of those on his ranch in Crawford, Texas.
October 9, 2024 at 8:45 am |
Dave,
Good Afternoon. It’s been a long time since I have heard from you. I trust you are well and that Helene left you and your family alone.
Broadly speaking, yes, I agree with almost everything you say! But here’s a couple of points you may find interesting.
Most heat pump systems in the UK don’t use house thermostats. They use weather compensation – this controls the heating by measuring the OUTSIDE temperature. This avoids a lot (but not all) of the cycling behaviour of conventional gas boiler based heating systems with room thermostats. Ideally, the compressor speed varies to match heating demand.
NC has much greater temperature extremes than the UK – and as you say it’s expensive (and wasteful) to build a heat pump system to match the most extreme conceivable temperatures. For UK climates, air source heat pumps now give ground source heat pumps a very close run for their money – their Seasonal COPs are very close and often above four. Check out this live monitoring site https://heatpumpmonitor.org
Best wishes
Michael
October 9, 2024 at 10:07 am
Interesting! I’d heard of variable speed heat pumps, but never encountered one in the wild. I had not heard of the term “weather compensation.”
A web search found this article.
October 9, 2024 at 10:12 am
Hello Again. In Europe, variable speed heat pumps dominate the domestic market. These are air-to-water systems that are mainly used for heating. They can modulate their output smoothly from around 100% down to 40%. For capacity factors less than 40% they have to cycle on and off. Their COPs and SCOPs are excellent easily exceeding what used to be the standard guess of “3”.
Best wishes
Michael
October 9, 2024 at 6:26 am |
Hi Michael,
My heat pump was oversized due to too high air exchange presumption in the heat loss calculation. However, despite being oversized I have improved the SCOP from an initial 3.59 to about 4.18 by lowering the maximum flow temperature from 50°C to 40°C. When it was briefly -20°C outside it was still keeping the property at a cozy 23°C whilst running very long almost continuously on hardly off cycles.
Although I could probably gain another 0.2 SCOP with a smaller heat pump it would likely have struggled to keep the house warm on the coldest days. The headline size ratings of heat pumps can be misleading as they are often rated at higher ambient temperatures than the coldest days where the power would be needed. At low ambient temperatures they won’t be running at their full headline power output.
The spreadsheet heat loss calculation I have includes the selected heat pump output profile against ambient temperature. The headline figure tends to match +10°C ambient and output falls by 20% or so at negative ambient temperatures.
Personally I’d rather have a comfortable capacity margin rather than faff around with electric heaters on cold days.
I’d also question the perspective of a marginal cost benefit calculation that predicts very long payback periods for the extra cost of a marginal change when the whole system cost and payback period when recalculated hardly changes for the same marginal change. For example, a larger battery for a PV plus battery and heat pump system may increase the overall system payback period from 5.7 to 6.2 years (a marginal extra wait) but if you only consider the cost of the extra battery and it’s marginal yearly saving improvement you end up with 20 years payback or something silly. Both can’t be right. It’s easy to lose perspective with such considerations.
October 9, 2024 at 8:52 am |
John
Good Morning. Firstly I am happy to hear how well your system is working out.
Yes, the headline or nameplate power can be misleading since different manufacturers have different criteria for naming. So names don’t match actual performance, especially in the cold.
And Yes, it’s nice to have a “comfortable” capacity margin. In a sense the article is about how big that margin could be and what happens when it’s exceeded. Obviously the best and simplest answer is to make things “just right” but sometimes there are difficult decisions to be made and I feel it’s important not to let the perfect be the enemy of the adequate.
And I agree with you 100% about marginal rates. They can be very highly deceptive.
Best wishes
Michael
October 9, 2024 at 7:19 am |
Very interesting. One thought is if there is desire for lower sound levels an undersized unit might not be right. As the quoted sound level is at 100% duty cycle. So a saller unit will be running at this sound level in milder times when people are more likely to be outside and hear it. Whereas a large unit ticking along at a lower duty will probably be the quite option? Just a thought.
October 9, 2024 at 8:28 am |
Joe, Good Morning,
You are quite right about the fact that larger unit would be running at a lower capacity factor and so might be quieter. But there are imponderables such as this and then there are Councils. The noise output is not specified as a function of power output and so the nominal sound output is all one has to go on.
Nothing is simple!
Best wishes
Michael
October 9, 2024 at 7:57 am |
You quoted the size of the heat pump but not the heat loss Econic determined. This is misleading as the Aerotherm plus ( and all heat pumps ) have different outputs depending on the flow temperature at design day. For example the 7kW Aerotherm plus has 8.8kW output with flow temp of 35 deg and 8.6 kW at flow temp of 40 deg for design day of -3 deg C. It is only with a flow temperature of 55 deg the 7kW Aerotherm plus has output of 7kW at -5 deg C.
October 9, 2024 at 8:23 am |
John, Good Morning,
Yes, indeed. The nameplate heating power of the heat pump can be misleading. For most manufacturers the number is (commonly) an overestimate of output at most temperatures, but as you point out, for Vaillant, it’s the other way around.
All installers are aware (Econic included) of the need to match output on design day with actual heat pump output and radiator emission. For all heat pumps the maximum heating power declines as the external temperature falls. In the spreadsheet you can model this as a simple linear trend.
Best wishes
Michael
October 9, 2024 at 9:04 am |
“I chose the example above because it is typical of many UK homes” oh my god Michael 😂😂. Average houses (mean or median) are a hell of a lot smaller than you think they are 🙂.
Average dwelling floor area is 94 sq m and average EPC is D. As Ofgem will tell you, “typical” (read mean) gas consumption is 12,000 kWh. So about half of the country needs a heat pump of just 4 kW – or less!
10+ kW installations really are outliers, way out on the right-hand tail.
October 9, 2024 at 9:14 am |
Dan,
Here is data for ASHPs from the Boiler Upgrade Scheme for August 2024
Mean capacity of installation (kW) 10.1
Median capacity of installation (kW) 9.0
Lower quartile capacity of installation (kW) 7.0
Upper quartile capacity of installation (kW) 12.0
So the example I gave is indeed typical of installations taking place in the UK now.
Best wishes
Michael
October 10, 2024 at 4:58 am |
Hmm… I strongly suspect this is a case of “lies, damn lies, and statistics”! Folks who apply for a subsidy on their heat-pump installation are — I’d guess — much more likely to be wealthy than poor, are living in larger than median housing, and (as Michael points out) are likely to be oversizing their heatpumps. The median household consumption of gas will depend (of course!) on the characteristics of the household… and a central tendency (whether a median or a mean) of a statistic which depends strongly on factors (such as household income or size) can be quite misleading. I live in NZ so have little interest in digging into the details but you might be interested to look at Ofgem’s 2013 (!) report on moving “beyond average consumption” in its policy analyses. If so… see https://www.ofgem.gov.uk/sites/default/files/docs/2013/06/beyond-average-consumption-summary-doc_updated-june13_0.pdf
Anyway: Michael’s study was illustrative. A full analysis would look at a *representative* range of households — perhaps using Ofgem’s 2013 breakdown into 12 different “consumer groups” with respect to their consumption of gas and electricity. The largest group is Archetype #9, 32% of the UK households, with an average of 16386kWh annual gas consumption and 3588kWh of electricity consumption. I’m idly wondering if the UK has made enough progress on home-insulation in the past decade that the total energy consumption for heating in this cohort has decreased significantly?
October 10, 2024 at 11:12 am
Clark, good afternoon. I trust you are well.
Yes, there is a bias in the distribution of people installing heat pumps: the people with the money to install heat pumps are better off and live in larger houses.
The more I have thought about it, the more I think that the kind of air-to-water (hydronic) system that I have just doesn’t make sense for smaller homes.The heating demands are low and the complexity of installation and siting just seem like impossible problems.
Your link to the Archetype categorisation document is fascinating: it feels like a pencil sketch of the entire nation!
Best wishes
Michael
October 9, 2024 at 11:51 am |
I’ve often thought oversizing was rife in the industry, and it’s comforting to see a smaller solution considered in detail. I have a gas boiler about 5 y old. I’m interested in a heating/cooling sytem utilising a heat pump, with the boiler initially retained to provide backup heating. Thanks so much for all your blogs; they form a comprehensive reference manual for the design.
Cheers
Terry
October 9, 2024 at 11:55 am |
Thanks Terry. Glad it makes sense to someone!
Using a gas boiler as a back-up is called a hybrid system and unfortunately isn’t eligible for the Boiler Upgrade Scheme bursary 😦
Best wishes
Michael
October 10, 2024 at 8:04 am |
Hi Michael. Fantastic article. Thank you so much for writing it and producing the calculator. I’m 100% aligned with your opinion on the matter and only read a similar article from Mario on Dodić Linkedin this week https://www.linkedin.com/pulse/optimal-heat-pump-design-main-challenge-mario-dodi%C4%87-tskff/?trackingId=8X%2FmfdKOSwOrSbCAsgVm%2Bg%3D%3D
I too struggle to accept that the current heat pump sizing is the correct approach given that everyone is actively overestimating the heat loss calculation or oversizing the HP.
October 10, 2024 at 11:27 am |
Alan,
Thank you for your kind words. I do think the whole sizing issue will gradually disappear as installers grow in experience and we all come to better understand the technology. In a way it’s growing pains!
Anyway: best wishes: Michael
October 11, 2024 at 2:49 pm |
My installer initially quoted for a 10 kW system. When I asked him to show me the MCS calculations it was borderline 7/10 kW. They had selected a 10kW unit ‘to be on the safe side.
By this time I had come across your excellent ‘Rule of Thumb’ which produced in my case the figure of 5.2 kW. I insisted on a 7 kW system which has now been running for over a year and has kept us perfectly warm (20 C) throughout the winter months.
The SCOP is 4.0 and no radiators were replaced, despite some being very elderly Type 10.
October 11, 2024 at 3:08 pm |
Dear Littlerichley,
Thank you very much for your comment which I find humbling. I am *so* happy your system is working and that you managed to avoid an oversizing error. I will take it as encouragement to keep going!
Best wishes
Michael
October 13, 2024 at 9:59 pm |
I have been following your blog for some time and find them very interesting. However this post contains a couple of assumptions that are not always true. First assumption that the heat pumps have a similar SCOP. The Arotherm 10kW pump has better SCOP that 7kW pump. The second issue was that providing back up heating would mean you wouldn’t qualify for the boiler upgrade system. However a very interesting article.
October 13, 2024 at 10:20 pm |
Good Evening
I am glad you finding the articles interesting. I write about what interests me, and after that it’s all just a hope that other people feel similarly. But whatever one writes, one is inaccurate to some extent, particularly if one tries to write compact articles.
Regarding the COP issue, indeed the 10 kW heat pump and the 7 kW have different characteristics. However I think this is what I would call a second-order effect: I think the performance of the heat pumps is “similar”. Also modelling the changing COP versus external temperature as a straight line is also crude. If you would like to develop the model further, then please be my guest – you can download the spreadsheet in the article and adjust it. Nonetheless, I feel the article conveys the point that the heat pumps would behave similarly until the smaller heat pump reached its capacity.
Regarding MCS, my understanding was that the installation qualified for the BUS, perhaps because the back-up heating was plumbed in. I’ll check and get back to you if I’m wrong.
Best wishes
Michael