Friends, the coldest day of the year affords a rare opportunity to find out the key number that describes the thermal performance of your dwelling: the Heat Transfer Coefficient (HTC).
This is particularly relevant if you heat your home with a gas boiler because, thrillingly, it also allows you to estimate the size of heat pump your dwelling will require when it’s time to switch.
The coldest day is probably still a couple of months away, and that gives you time to prepare and practice for your day of measurement.
Let me explain.
The Coldest Day?
Let me begin with the profound philosophical question: “How do we know which day is going to be the coldest?”. As the Zen master said “Even a very cold day may be followed by colder days.”
Fortunately, we don’t the need the very coldest day of the day – any reasonably cold day will do, and the weather forecast should alert you to its arrival. Ideally it would be a day with an average temperature close to 0 °C, perhaps with a nighttime minimum temperature well below 0 °C.
And seeing how the results compare on a couple of similarly cold days will help you assess the likely uncertainty in your estimates.
A Very Cold day
On this cold day you need to use your electricity and gas appliances as you would normally so that your home is as warm as you would like.
Then you need to read your electricity and gas meters before the coldest night – and then at exactly the same time the next day. Alternatively a smart meter might well give you the information more conveniently.
Finally you need to know the average temperature inside and outside your home.
I’ll explain how to do the calculation below but with these readings you can estimate the Heat Transfer Coefficient (HTC) for your dwelling and – when the time comes to change – the size of heat pump you require.
The General Idea
Let’s say you used 20 kWh of electricity and 100 kWh of gas over a period of 24 hours. Then a simple first guess would be that the total energy used for heating was 120 kWh. Over 24 hours this would correspond to an average heating power of (120 kWh)/(24 h) = 5 kW. We’ll make a more sophisticated estimate below but this would be a good first estimate of the size of heat pump you require.
If the internal temperature through the day was 20 °C and the average outside temperature was 0 °C, then you can estimate the HTC by dividing the average heating power (5 kW) by 20 °C i.e. 5 kW/20 °C = 0.25 kW/°C.
A more sophisticated estimate
Electricity. Unless you are charging large batteries or directly heating hot water with an immersion heater, all the electricity you use – for televisions, lighting etc – ends up heating your home. So the 20 kWh of electricity used would all end up as heat.
Gas. In a typical boiler, 15% of the energy in the gas that went through the meter is lost out the flue. In older boilers losses could be as much as 25%. If you don’t know better, a good first estimate would be that your boiler efficiency was 85%. So if 100 kWh of gas was metered, I would estimate that only 85 kWh actually entered the dwelling.
Cooking with gas. The heating power of gas used for cooking is generally small (a few kWh/day) and most of the heat ends up in the home any way.
Domestic Hot water. Gas or electricity used to heat water doesn’t generally heat the home (much) and needs to be subtracted from the estimate of heat supplied to the house. The industry guideline is that each adult uses about 3 kWh/day of hot water, so if there are two adults in the house you need to subtract 2 x 3 = 6 kWh/day from the estimate of gas used for heating the house.
People. People are actually a source of heat, releasing around 2.4 kWh/day. If there were two adults in the dwelling all the time then add 2 x 2.4 = 4.8 kWh to the heating energy.
So the total heating in the dwelling would be 20 + 85 – 6 + 4.8 = 103.8 ± 5 kWh. The uncertainties are such that this can be conveniently rounded to 104 kWh rather than 120 kWh in the simple estimate.
So over 24 hours this allows you to estimate of the size of heat pump you require as being (104 kWh)/(24 h) = 4.3 kW.
If the internal temperature through the day was 20 °C and the average outside temperature was 0 °C, then the HTC is estimated as 4.3 kW/20 °C = 0.22 kW/°C.
If you wanted the heat pump to keep your home at 20 °C when it was (say) -5 °C outside, then you can use the HTC to estimate the size of heat pump required. You multiply the temperature difference (20 – (-5)) = 25 °C) by the HTC (0.22 kW/°C) to give 25 °C x 0.22 kW/°C = 5.4 kW.
I have prepared a spreadsheet that does the calculations for you:
- Coldest Day Spreadsheet [Updated on 24 March 2023]
Why You Need a Cold Day
You can make these measurements on any day of the year, but except on the coldest days, the uncertainty in the estimate can be large.
By making the measurements on a cold day, the main heating component – the gas consumption – can be estimated modestly well, and all the corrections are relatively small. My guess is that the answers should be within about 10% of the right answer.
Reading the Meters
If you have a smart meter with an in-home display, then one of the settings will tell you how much energy you have used in the last day. Typically, they show data for gas and electricity separately, each for a 24-hour period starting at midnight. If the weather stays cold for two days, it might be better to record energy usage over a 48 period so as to include a complete cold night.
If you don’t have a smart meter, then you will have to find out where your energy meters are in your home and read them manually. If you don’t know how to read an energy meter there is help available from:
- Citizen’s Advice Bureau (Gas and Electricity Meters)
- British Gas (Electricity Meters)
- British Gas (Gas Meters)
But there is one difference between reading the meter for an energy company and reading it for yourself. When reading the meter for the energy company they tell you to miss off the last digits. This is because they want to minimise the chance of mis-reading and transcription errors. And they know that what you don’t pay for this month you will pay for next month!
But there is information in these digits which can be useful, especially if your usage is low. So record all the digits from your gas meter.
Click image for a larger version. The left-hand image shows a gas meter reading in cubic metres and the right-hand image shows a gas meter reading in cubic feet.
Gas meters record your gas usage by measuring the volume of gas passing through them in cubic metres or cubic feet. To estimate the energy contained in that gas you need to subtract the volume readings made at the start and end of your chosen 24-hour period, and then multiply by a factor which tells you the energy content per unit volume of the gas.
Click image for a larger version. Spreadsheet excerpt showing how to subtract two readings to obtain the volume of gas used in one day, and multiply them by the energy density to find the energy contained in the gas that flowed through the meter.
Older gas meters sometimes confusingly read in units of hundreds of cubic feet rather than cubic feet. An example of this is given in the illustration above. If you are unsure you can check that you have the right units because for any reasonable home heated primarily by gas, the gas used on the coldest day of the year will be somewhere between 10 kWh (a well-insulated flat) and 200 kWh (a large poorly insulated house).
Fortunately electricity meters read directly in kWh.
A spreadsheet that does the calculations for you can be downloaded here:
Click for larger version. Graphic showing the spreadsheet that will do the calculations for you.
Temperatures
Reading your meters can be tricky, but working the average temperature inside and outside your house can be trickier.
The best way to do this is to measure it yourself with thermometers and weather stations. For most people that’s not possible.
If you don’t have an internal thermometer, then I have been told that the average household temperature is likely to be approximately 2 °C colder than the thermostat setting. So if your thermostat is set for 20 °C, then the average temperature of the dwelling is likely to be around 18 °C.
To estimate the average external temperature you might try this web site which allows you view historical weather data in your location. This link is for London, but you can choose other locations.
Alternatively use the Weather Underground’s Wundermap to find a local weather station. You can zoom in to a local level and click on an individual weather station and then its weather station ID to get its local daily and weekly average temperatures.
Last thoughts
Friends, the essential and expensive energy which flows into and out of our homes is sadly invisible. And this makes it difficult to assess the thermal properties of your home.
But the coldest days of the year afford us an opportunity to assess the thermal properties of a home that only comes about on a few days a year.
I urge you to get ready for when the cold days arrive – perhaps by practicing on less cold days – and then you will be able to obtain valuable information about your home.
Good luck!
Protons for Breakfast 
November 7, 2022 at 3:00 pm |
Fascinating!
November 7, 2022 at 4:24 pm |
I am happy you found it so. Thanks for stopping by. M
November 7, 2022 at 3:38 pm |
Interesting stuff, as ever, Michael – thanks!
I declare my interest – I am such a gas boiler owner who aims to use it to do exactly as you describe, to gather information that would reliably inform the specification of a heat pump that could replace it.
As you say, the analysis is a matter of balancing energy input (including what’s pumped in) against energy lost but, as we once discussed, there can be more to the losses than the temperature difference, inside – out.
If I recall our discussion correctly, you’d looked (in vain) for evidence of wind-speed dependence in your own energy consumption, while I’d found that far and away the biggest peak in the energy consumption in my house was a period of several days of sustained ~30 mph winds, from what one would expect to be the “worst” direction, which is when the wind blows face on to the terrace which includes the house. To boot, we benefit from uninterrupted sea views in that direction 😉
In short, but in a very building-dependent way, significant energy loss _can_ arise from convection – the exchange of expensively heated air and cool outdoor air.
Relatedly, a few days ago I stumbled across this old article https://passivehouseplus.ie/magazine/feature/windy-or-cold-weather-when-does-heating-demand-peak
which also makes for interesting reading.
They go into some examples where excellent thermal insulation was seriously compromised by these effects.
Best wishes
November 7, 2022 at 4:33 pm |
Simon,
Good Afternoon. The gist of the article is that one finds out experimentally how much heat one uses on an extreme day. For most people, the extreme is simply one related to temperature. For your extremely exposed location the heating demand extreme may come at some combination of cold and windy.
Note that I don’t think the additional heat loss in a windy state is caused by air speed across the exterior surfaces. I think heat losses at those surfaces are essential instantaneous compared with the rate of heat transfer through a brick wall.
Rather, the heat loss likely arises from air flow through the interstices of your home driven by the extreme ΔP across your house. Air tightness tests are conducted at ΔP = 50 Pa, but there is large fact – perhaps as much as 30 (if I remember correctly) that relates the number of air changes per hour at ΔP = 50 Pa to what one might expect in a normal domestic setting. On a seafront, it could be your number of air changes per hour is very high.
But whatever it is, I expect that a measurement of gas consumption on a cold and windy day will allow you to estimate it. Indeed comparing cold and still with cold and windy would be a great experiment!
All the best
Michael
November 7, 2022 at 5:03 pm
I’m not sure if you followed that link, but it was, of course, not about my own house but about the housing (and also commercial building) stock in Ireland (rather than the UK), and was making a very general point about the two mechanisms of heat loss.
Two terms are used (and distinguished)
– “airtight”, which I understand to be quantified in a blower test, by increasing the pressure indoors by 50 Pa, and
– “windtight”, which sounds like it should be meaningful, but I would struggle to suggest how to quantify it. (I imagine that the pattern of air flowing over a building creates pressure differentials on windward and leeward sides, Bernoulli effects in chimneys, etc. I suppose that the pattern of winds makes some of the leaks that spoil airtightness more important than others.)
As you mentioned, smart meters can make these things easier to investigate. Indeed, a “simple” approach is to review daily combined energy consumption over the year (after the fact), to home in on the peaks, and ask – what caused that? before trawling through weather records etc in search of a plausible story.
All good fun …
November 7, 2022 at 3:52 pm |
PS Re-reading the article I linked to, I see there is an interesting hint, which I shall make more explicit:
if the maximum demand for heat is when it is windy but not sub-zero, I might expect an air source heat pump to come into its own: wouldn’t the wind reduce the tendency for the external unit to get iced up when it’s trying to pump as much heat as possible into the house?
November 7, 2022 at 4:37 pm |
I don’t think windiness will particularly affect ASHP performance. Why? because the air flow past the evaporator is already very fast. A bit of extra wind will not make much odds.
The main factors affecting condensation of water – and ice – are around the water content of the air. I analysed this here: https://protonsforbreakfast.wordpress.com/2021/08/30/heat-pumps-power-noise-and-condensation/
Personally I wouldn’t worry about the de-icing issue. There is nothing you can do about it! and it only affects COP on a few days a year. Better to look for better performance throughout most of the year.
Best wishes
M
December 31, 2022 at 10:20 am |
Thank you for this, very interesting and I shall work through it.
I’ve noticed a copy/paste typo in cell B22. The title I believe should be ‘Gas Traditional Meter (Cubic Metres)’ rather than cubic feet.
December 31, 2022 at 10:33 am |
John, Thank you very much for the typo which is now corrected. Happy New Year and all that. M
January 19, 2023 at 5:56 pm |
Hi Michael, two quick points.
There appears to be an error in G16 formula, not coping with ‘Smart’ meters.
Second – I always ask when someone states their gas use for a very cold day “and were you actually comfortable throughout the day/evening?”. This is because I always accepted that my old gas boiler would provide comfort for the hours it was running, but much less comfort outside the run time (maybe running only 6-7 of 16-17 waking hours). Many people will probably feel the same. I have an ASHP now, running ‘low and slow’ with a very pleasant all-day temp. The quality of life improvement is very significant. In summary the cold days calculation is still valid of course, but how many people avoided boosting the gas when it got cold and still used > 100kWh?
January 20, 2023 at 12:00 am |
Bob, Good Evening.
You make a fair point. There is no substitute for accurate measurements, and the way people live in their homes is complicated.
The reason I wrote the article about using cold days is that in building science, even an *indication* of teh approximate value of the heat transfer coefficient is valuable.
I’ll check out the spreadsheet. Thanks
Michael
January 19, 2023 at 6:03 pm |
This article was some suggested reading you gave me for considering what size heat pump I may need to replace a gas boiler and it is very helpful. Something I found useful for accessing energy data for my house is the free Uswitch app I’ve been using for around 3 months, which logs hourly gas and electricity consumption and allows review of historical data. I have my own local outside temperature data so I could go back to Dec 15th last year when our outside temperature was -8 degC and use the Uswitch app to get the total gas and electricity consumption for that day. There is also free historical temperature data for UK locations if you don’t have your own data eg. see http://www.timeanddate.com/weather
Looking at the Uswitch data on an houry basis also gives potential insight into the performance of the gas boiler during the course of the day and the impact of how the boiler is best programmed. Makes me wonder whether I can use this data to reduce the energy total for the day – if you have any suggested reading for making the most of a gas boiler I’ll follow that up! I suspect opitimising the use of gas boilers is a priority not just for reducing current CO2 & energy costs, but for helping to avoid the over-specifiying of air source heat pump replacements.
January 19, 2023 at 7:04 pm |
Gordon, well done: what fantastic resources. I use Time and Data a lot but I hadn’t realised they had weather data.
Regarding your boiler, the standard suggestion is to try lowering the flow temperature of the water flowing through the radiators.
If your boiler heats water to 70 °C and pumps it around the radiators then it will transfer too much heating power to the house and the thermostat will switch off the boiler for 20 minutes.
If you lower flow temperature to (say) 65 °C the boiler will operate a little more efficiently and stay on for a bit longer.
Try some experiments to keep lowering the flow temperature until the boiler stays on most of the time. If you can get your flow temperature down to 50 °C while your house stays warm, then you will have no difficulty using a heat pump.
Good luck! M
January 18, 2024 at 10:30 am |
There is something quite fundamental here that I am struggling to understand. Why are you averaging heat used (by gas) over 24 hours, when the boiler is not used over 24 hours?
Would it be more accurate to concentrate on the hours the gas heating is timed to be on, and the house temperature has stabilised?
If I look at my own data. I used about 57 kWh of gas yesterday, over about 14 hours, outside temp. about 0 degrees C all day.
If I divide by 24 I get about 2 kWh. If I divide by 14 I get 4 kWh. If I look at my smart meter I can see my heating was “actually” using about 4 kWh per hour to keep my house warm. So a heat pump of 2 kW would not seem to be sufficient, whereas a 4 kW one would.
Have I misunderstood something?
January 18, 2024 at 3:44 pm |
Mark C, Good Afternooon,
No. I think you have understood things perfectly but I may have miscommunicated. My assumption is that the house temperature is more or less stable through the day. As you say, if you don’t heat the home for a significant fraction of the day, then the gas consumption will appear anomalously low and the estimated size of heat pump would be underestimated.
With a heat pump, it would be recommended to either leave it running continuously, or to only have a slight ‘set back’ temperature – perhaps just a couple of °C cooler than normal. Why? Suppose a 4 kW is happily heating your home. At this point, all the heat you put into your home will flow straight out. If the house were (say) 10 °C colder then you might want it to warm up relatively quickly to normal temperature. If 4 kW were its maximum heat output, then it would only reach your desired temperature rather slowly. So changing the temperature of a house *quickly* is something that a gas boiler can do, but a heat pump can’t.
Sorry if the article wasn’t clear – but I think you understanding is clear.
Best wishes
Michael
January 19, 2024 at 3:28 pm
Thanks for clarifying Michael. Yes, my gas boiler uses about 8 kWh for the first hour, which as you say a heat pump is not really designed for.
Regards
Mark
November 23, 2024 at 9:42 pm |
We’ve just had a cold snap and it reminded of your blog What to do on the coldest day of the year. On 20/11/24 our ave Living Room temp was 19.5 degC and the ave Outside temp was 0.4 degC (a min of -3.4 degC). Gas was 95 kWh for the day, which allowing for 90% bolier efficiency would be a heat input of 85 kWh. Electricity was 7kWh which makes a total heat input of 92 kWh. So delta T = 19.1 and the required heat pump size would be around
(92/24) x (25/19.1) = 5.02 kW
Octopus have recently undertaken a heat survey of the property (1970 4 bed detached house) with total room heat losses coming out at 9.445 kW leading them to quote me for a 11 kW heat pump! The current radiators provide 7.076 kW and hence it’s proposed to replace several radiators with larger outputs.
I’ve had some thoughts about this and will send you a separate note.
PS. Our latest annual gas consumption is just over 16,000 kWh which would suggest a heat pump size of around 5.5 kW using your 2900 rule. I can’t remember whether this rule takes into account boiler efficiency, which is 90%.
November 23, 2024 at 10:14 pm |
Gordon, Good Evening. And thanks for that report on your measurements. Your comment makes it clear to me again just how valuable the cold weather can be.
I agree with your calculations: The “Rule of Thumb” estimates boiler efficiency at 90% and 5.5 kW seems like a much better estimate that 11 kW!
What to do? Well I would recommend that you don’t allow an 11 kW heat pump to be installed. It’s a massive unit and it would only rarely be able to operate continuously. A Vaillant Arotherm plus 5 kW model actually outputs more than 5 kW – especially if the flow temperature can be kept low. Get a 7 kW model if you want to be cautious, or allow for a future user to extend or operate at higher room temperatures. But 11 kW is crazy.
Are you with Octopus Energy already?
Best wishes
Michael
November 24, 2024 at 3:53 pm
Thanks for your prompt and helpful reply. I am with Octopus and moved to them about a year ago in order to secure their 15p PV export rate. As well as offering a very good rate, export payments are paid monthy and directly credit the import electricity account. Something not often mentioned is that swapping energy import suppliers needs to take into account what export rates they offer. Sometimes there may be a case to only move the gas supply – Octopus have advised they only require electricity import to be from them to offer their 15p export rate. My experience of changing energy companies for export has been very poor (due to DNO delays) compared to that for changing for import. It will be intereting to see what will happen to the Octopus import rate of 15p when its reviewed in March/April next year.
The free heat loss survey undertaken by Octopus was offered as a precursor to taking advantage of the £7,500 grant currently on offer for gas to heat pump conversion. Whilst the detailed radiator assessment was very useful, I wasn’t able to get very much clarification on what had been taken into account in arriving at the calculated figures. I’ve detailed some of these below and hope they will be useful comments for others.
The Octopus assessment made reference to the EPC for our house undertaken 9/8/22, which is out of date because we have had PV panels and additional loft insulation installed since that date. Nevertheless they used the EPC data and out of date metered energy consumption figures to compare with the estimated energy consumption of the proposed Daikin 11 kW heat pump, which is a ratio of 3.4 to 3.6 lower. Running at a flow temperature of 50degC, the tabulated SCP was given as 3.56.
I would be happy to install a heat pump even though there’s unlikely to be a saving on running costs for our current gas and electricty unit tariffs of 5.93p gas / 25.12 p, a ratio of 4.22. However, I don’t want an oversized heat pump. The quote came in at £12,060 leaving me a balance of £4,060 to pay. That includes a replacement Daikin hot water cylinder and several radiator upgrades. However, I have run our central heating at 50 degC flow temperature and found that there is adquate hot water for two people from our MegaFlo 170. Space heating was only a challenge in the Living Rooms & Utility Rooms both of which have 3 external walls, these accounting for 34% of the estimated tabulated total heat loss. However this was only when outside temperatures approached subzero. So a replacement radiator for the Living Room is needed, along with one for the Utility Room but not in our experience elsewhere, out of a total recommended replacement of 7.
Given the large discrepancy between current energy consumption and estimated heat losses for our house, there appears to be an inherent flaw in the overall assessment being undertaken for determining the size of a proposed heat pump installation – this is something you have been addressing for several years.
Perhaps the biggest flaw is likely to be in delta T. Our local Woodford Climate Station shows min temp data over 1991 to 2020. For Dec ave min = 2.73 degC, 80% = 1.15 degC, 20% = -0.68 degC. For Jan ave min = 1.07 degC, 80% = 1.88 degC, 20% = 0.09degC. So Octopus using an external temperature of -2.9 degC as an all day temperature for heat loss calculations will be overstating the delta T.
A second issue, again with delta T, is whether all external walls are equal. Our utility room, kitchen, and study all have a nominal external wall with our double garage. I monitor the garage temperature which on 21/11/24 was 6.4 degC when the outside temperature was -3.4 degC which represents a significant overestimate of delta of 9.8 degC. On the same day & time, our hallway shares an outside wall with a glazed porch, and the temperature was 5.4 degC, with a delta T overestimate of 8.8 degC. On the same day & time, our loft temperature was -1.6 degC with a delta overestimate of 1.8 degC, but I don’t know how heat loss into the loft is taken into account.
A third issue is around heat gain. Our house is due-South facing, with the most used rooms (kichen, living room, and master bedroom having a total glazed area of around 12m2 all facing due South. I assume that the amount of solar energy throughout the year can be calculated as for PV, taking into account the vertical face of the glazing and the daily positioning of the sun through out the year. However, I don’t know how to assess what temperature rise would result from the gain.
My final thought concerns the impact of the thermal lining we now have on all curtains. This is not taken into account in the EPC or heat loss assessment. This is a benefit beyond preventing draughts, but I don’t know how to assess any further impact on heat loss.
Apologies for the long contribution. I’ve tried to keep it brief.
November 25, 2024 at 11:21 am
Gordon, Good Morning
Your “brief” comment is 850 words long! Now you know how I feel when writing!
I have a couple of comments.
1. Regarding the poor survey, Yes, mis-categorised external walls can give rise to very large errors. Similarly, over estimation of the rate of air changes can also cause large errors. Good surveyors are aware of these pitfalls and will try to reconcile their survey with actual energy readings.
2. Solar gain is difficult to account for, but is not especially relevant to calculating the maximum heat loss – this is likely to occur in midwinter on days when solar gain is zero. It is however an issue for temperature control, particularly in autumn and spring. Most heat pump controls deliver heat based on the external temperature and so this can lead to overheating when then solar gain is significant. But I think all control systems will have some way of addressing this issue.
3. Insulation on curtains is probably not relevant unless the curtains are closed.
4. Using an external temperature of -3 °C and an internal temperature of 23 °C is standard practice for designing systems. It means installations can cope with the coldest likely temperatures, and with higher internal temperatures. If you pay all the money for the installation then you can install as you see fit, but if you want £7,500 of subsidy then it’s part of a national infrastructure upgrade, and the next owner of the house – or you when you are older – may want higher temperatures.
5. Regarding the economics of switching and export rates, I honestly don’t know how Octopus make money. Presumably they have a grand plan to capture the market and then make things slowly worse: it seems to be standard operating practice for big businesses. But while these good rates are available, they are hard to resist.
Best wishes
Michael
November 25, 2024 at 2:27 pm
Thanks for your speedy reply. To keep this shortish, I’d like to comment on your comments 1,2,& 3.
Best wishes,
Gordon
January 10, 2025 at 11:31 am |
The latest cold weather has given us -6.5 degC locally and that has prompted me to look further at refining the use of gas consumption for predicting heat pump size, a common criticism being that gas consumption may well not be based on full occupancy. The MCERTS approach associated with grant eligibilty is to assume that all habitable rooms have rads that are in use and that the hot water supply will handle full occupancy with any grant-supported heat pump installation.
To take into account rooms not in use, the m2 floor area of rooms not in use could used to estimate the additional heat required, given the known gas consumption for the m2 floor area that is in use. For our house we only regularly use and heat one of the bedrooms. The unused bedrooms account for 17% of the total house floor area and will likely increase total gas consumption by around that amount. As it happens I do have an MCERTS heat loss assessment for our house and that indicated a 15% of the total calculated heat loss for those unused rooms. The EPC will give total floor area.
To take into account hotwater requirement, the gas consumption over a period when no heating is required can be allocated to hot water consumption if no gas is used for cooking. For our house with 2 people (for August) that amounted to 4.4 kWh / day per person, somewhat higher than the industry guide of 3 kWh / day. This may well reflect a preference for baths rather tham showers.