I've seen a couple of ways of "estimating" heat pump size: one, by dividing yearly gas kWH by 2900; another, somewhat more complicated, using Heating Degree Day data. It got me to thinking about how close the outputs of these might actually be.
So, for example, here is my 2024 gas data: Heating 8472kWh; hot water 2221.2. Although the sum of these two numbers is spot-on accurate, the split is somewhat extrapolated as my gas usage doesn't actually break out what is used for heating, hot water and cooking. However, in July and August there is no heating and the gas consumption for hot water and cooking are at different times so I can take these and extrapolate them for the other months giving me a realistic monthly breakdown:
This gives a yearly heating and hot water total of 10492 kWh which would give an estimated ASHP size of 10492 / 2900 =3.618kW
I also created a HDD calc:
This shows an estimated ASHP of 3.5kW. It also shows that there should have been some heating usage over summer which shows how bad it was! The hot water load is only 3kW and given the ASHP is either heating or hot water(-ing) then hot water doesn't increase the size. Incidentally, I calculated the load by taking the yearly figure / 365 / 2 because it is on twice a day with around the same tank regeneration time.
Interestingly, they both arrive at similar figures. Both Octopus and Heat Geek have undertaken a heat loss assessment and arrived at very similar figure: 8.3kW and thus have recommended a Cosy 9 and Vaillant 7KW (because it can rate up to just below the calculated loss) respectively.
Both these models are, minimally, double the guesstimate so it got me to thinking.
What I think is often missing from these "double-check", "Ballpark", "etc etc" approaches, and which may give misleading views, is the actual running use vs a design use. Octopus and Heat Geek, and I guess others, design for a 21c (or mixed temp) model. My figures are from running the heating at midnight to 06:30 18c; 06:30 to 16:30 19c; and 16:30 to 23:00 20c. How much extra gas usage would accrue from, say, running at a day/evening temp of 20c and 21c or even a day/evening of 21c and setback of 18c?
Additionally, it made me think about the heat loss tools used and their accuracy. To generate an "accurate enough" heat loss the tools have to be simple and easy to use for a typical engineer otherwise they just wouldn't be cost effective. My expectation would be that calculated heat loss would tend towards pessimism, i.e. over value the heat loss. To get an accurate heat loss value would probably require weeks worth of data under controlled conditions using all sorts of tools and instruments.
Out of interest, I'm in the UK midlands, my house is a Finnish timber frame with wall insulation, floor insulation, 150mm loft insulation and triple glazed windows (0.5u) throughout. It typically loses approximately 2c of temperature in the room with the thermostat between midnight and 06:30 (in other words, the heating very, very rarely comes on with a 2c setback temperature.) The heat loss calculation was done assuming that 300mm loft insulation will be in place (which it will!)
the heat loss calculation is definitly an estimate. For starters the u values of the materials and the air changes are just estimated. For a well insulated house the setback at night makes little difference. From what you say when the temperature is lower at night on average it is 1C lower during the night, so all you’re saving is what the difference is between heating the house to 19 and 20 degrees. Depending on the outside temperature I’d think that is at most 10% over a third of the day, call it 3%, so you can ignore it. If your house was loosing lots of heat you’d save more.
When the house is well insulated, the air changes that they base the heat loss calculation on is a much bigger part. MCS I think advises a rate of 2, which is far too high and is most likely the source of the difference. The finish timber house is probably also better insulated and air tighter than your typical uk newish build brick house.
Now that I have the heat pump installed, the real world usage is closest to the calculation from gas usage (7000kwh/year => 2.4kw heat loss), then any of the heat loss calculations.
My figures are from running the heating at midnight to 06:30 18c; 06:30 to 16:30 19c; and 16:30 to 23:00 20c. How much extra gas usage would accrue from, say, running at a day/evening temp of 20c and 21c or even a day/evening of 21c and setback of 18c?
Heat loss (and therefore the energy you need to supply) is proportional to (house temp - OAT). Note that the house continues to lose heat even if the heating is switched off, and that heat has to be replaced when you switch it back on. This is why timed operation doesnt save anything like as much as the manufacturers of smart thermostats would have us believe even with a boiler (with a heat pump timed operation may cost more in some circumstances).
Average OAT over a year is about 7
So very roughly a house kept all day at 20 will need 2/(20-7) more energy, about 15%, more than one kept at 18. The percentage is actually a bit more than that because there is a fixed contribution to heating from the occupants (about 100W each) and electrical appliances (typically 400W).
If you switch the heating off for say 12hrs in a house kept at 20, and it cools to (say) 16, then the average temperature over 24hrs is a bit less than 19, so doing this saves 10-15%.
Additionally, it made me think about the heat loss tools used and their accuracy. To generate an "accurate enough" heat loss the tools have to be simple and easy to use for a typical engineer otherwise they just wouldn't be cost effective. My expectation would be that calculated heat loss would tend towards pessimism, i.e. over value the heat loss. To get an accurate heat loss value would probably require weeks worth of data under controlled conditions using all sorts of tools and instruments.
The standard 'MCS' heat loss calculation based on the fabric makes lots of assumptions (or rather the user does) and this can lead to significant inaccuracies (almost always oversizing). I had 2, 3 hour surveys done early on in my journey, both were a factor of 2 out (over) relative to the figure determined by plotting gas consumption in various ways from 2 years worth of half hourly meter data.
By the time I actually purchased the industry seemed to have matured somewhat so I ended up with a heat pump that is close in capacity to the figure one would expect from the gas consumption. In operation its right-sized so Im jolly glad I did the gas consumption figures.
There has been much debate about this on the forum. My personal recommendation is use all the data you have to get to the closest you can and as a minimum sense check any fabric spreadsheet against annual gas consumption, building in some reasonable assumptions.
I guess my point is that you have to be careful with these estimates for confirmation as it could well lead to some false assumptions, e.g. my heat pump is going to be oversized. I have seen a number of statements in different places (do I correctly recall a whole YouTube video??) along the lines of "if you want to get a rough idea just use your yearly gas usage and divide by 2900". Really, you aren't comparing apples unless the data you are using for the comparison/confirmation is based on a heating schedule that mirrors the design criteria.
My Heat Geek definitely said that the air loss used in the calculation was overstated, which was another pointer towards a 7kW model than the 10kW one.
@andrewj Even using actual gas usage, it is necessary to make assumptions about the boiler efficiency and the proportion of gas used for DHW and cooking (although it could be argued that these last two are within the 'heat envelope').
I'm presently attempting to measure my heat loss one room at a time using a thermostatically-controlled fan heater and measuring the electical energy input, but it's not as straightforward as I originally thought - I keep uncovering more potential sources of error.
Even using actual gas usage, it is necessary to make assumptions about the boiler efficiency and the proportion of gas used for DHW and cooking (although it could be argued that these last two are within the 'heat envelope').
As you say you have to make assumptions about boiler efficiency, probably best based on boiler age and whether it actually condenses or not.
Cooking is almost negligible almost certainly unless you live in a passivhaus or run a restaurant. DHW is part of the capacity requirement anyway, so in a sense the fact its included isnt necessarily a problem. You could separate both out by subtracting the summer values, unless like me your summer DHW was entirely provided by solar.
I guess my point is that you have to be careful with these estimates for confirmation as it could well lead to some false assumptions, e.g. my heat pump is going to be oversized. I have seen a number of statements in different places (do I correctly recall a whole YouTube video??) along the lines of "if you want to get a rough idea just use your yearly gas usage and divide by 2900". Really, you aren't comparing apples unless the data you are using for the comparison/confirmation is based on a heating schedule that mirrors the design criteria.
My Heat Geek definitely said that the air loss used in the calculation was overstated, which was another pointer towards a 7kW model than the 10kW one.
Unfortunately there is, as yet, no simple perfect way to determine heat loss that is universally applicable and that can be used without taking some care. I have seen various divisors to get to house loss from gas consumption and personally wouldn't like to be any more precise that 2000-3000 given how crude this method is (Im assuming that the house remains warm, even if its not heated, for most of the day). I tend to regard total gas consumption as a 'sense check', but an important one (with Oil, unless you have logging, its all you have!).
For my own determination I took 2 years worth of smart meter data, plotted daily consumption against degree days (offset by about half a day to account crudely for the thermal mass), and projected that. I also plotted average consumption averaged over 3, 6, 12 hours against a timeline, and eyeballed the peaks. I attach the data in case anyone is interested Lastly I turned my boiler down to the minimum it could go to which was 8.5kW. And of course I did my own version of the spreadsheet method. All the measurements from the gas came to the conclusion that the loss was ~7kW. The spreadsheet, done by me, got initially to 10.5kW but then I was tipped off that the default ACH values are extremely 'generous'. If I put 0.5ACH in to fit the data to the measurement, the spreadsheet gets to 7.5kW. Two out of the several installers I contacted didn't need convincing that 0.5ACH was a sensible value and I have since heard a couple of installers say that, whenever they have done measurements of ACH, they have never encountered a value >1. Clearly a lot of this is a matter of judgement in the absence of solid data, which is why i suggest to collect all the data you can!
I have a hunch that clever analysis of half hourly smart meter readings vs OAT could yield a wealth of data. Ideally you would also have IAT, but I'm guessing you can come close to inferring that, to sufficient accuracy, from the dynamics with a sufficiently intelligent data mining algorithm. I'm also guessing Octopus are working on that!
@jamespa That's a good job James. I expect many people won't go to that level though and just scratch their head over why their billed usage / 2900 doesn't seem to match what they are being told and what that implies. These days it is very easy to jump to a conclusion of being overcharged or over specified particularly as that heat loss figure impacts more than just the choice of heat pump size.
Personally, my money would be on a heat loss in my house of < 8kW but > 6kW because of the air change value being overstated - that's still way more than the sense-check value. So it really means that I (and others) would need to have a lot more data and more sophisticated analysis to arrive at a sense-check value as you have done so well. In other words, relying on a 2900 figure is pretty meaningless.
I've seen a couple of ways of "estimating" heat pump size: one, by dividing yearly gas kWH by 2900; another, somewhat more complicated, using Heating Degree Day data. It got me to thinking about how close the outputs of these might actually be.
So, for example, here is my 2024 gas data: Heating 8472kWh; hot water 2221.2. Although the sum of these two numbers is spot-on accurate, the split is somewhat extrapolated as my gas usage doesn't actually break out what is used for heating, hot water and cooking. However, in July and August there is no heating and the gas consumption for hot water and cooking are at different times so I can take these and extrapolate them for the other months giving me a realistic monthly breakdown:
This gives a yearly heating and hot water total of 10492 kWh which would give an estimated ASHP size of 10492 / 2900 =3.618kW
I also created a HDD calc:
This shows an estimated ASHP of 3.5kW. It also shows that there should have been some heating usage over summer which shows how bad it was! The hot water load is only 3kW and given the ASHP is either heating or hot water(-ing) then hot water doesn't increase the size. Incidentally, I calculated the load by taking the yearly figure / 365 / 2 because it is on twice a day with around the same tank regeneration time.
Interestingly, they both arrive at similar figures. Both Octopus and Heat Geek have undertaken a heat loss assessment and arrived at very similar figure: 8.3kW and thus have recommended a Cosy 9 and Vaillant 7KW (because it can rate up to just below the calculated loss) respectively.
Both these models are, minimally, double the guesstimate so it got me to thinking.
What I think is often missing from these "double-check", "Ballpark", "etc etc" approaches, and which may give misleading views, is the actual running use vs a design use. Octopus and Heat Geek, and I guess others, design for a 21c (or mixed temp) model. My figures are from running the heating at midnight to 06:30 18c; 06:30 to 16:30 19c; and 16:30 to 23:00 20c. How much extra gas usage would accrue from, say, running at a day/evening temp of 20c and 21c or even a day/evening of 21c and setback of 18c?
Additionally, it made me think about the heat loss tools used and their accuracy. To generate an "accurate enough" heat loss the tools have to be simple and easy to use for a typical engineer otherwise they just wouldn't be cost effective. My expectation would be that calculated heat loss would tend towards pessimism, i.e. over value the heat loss. To get an accurate heat loss value would probably require weeks worth of data under controlled conditions using all sorts of tools and instruments.
Out of interest, I'm in the UK midlands, my house is a Finnish timber frame with wall insulation, floor insulation, 150mm loft insulation and triple glazed windows (0.5u) throughout. It typically loses approximately 2c of temperature in the room with the thermostat between midnight and 06:30 (in other words, the heating very, very rarely comes on with a 2c setback temperature.) The heat loss calculation was done assuming that 300mm loft insulation will be in place (which it will!)
Hi James,
The actual values used for MCS are the room design temperature, the insulative values (u-values) for the property as well as the unique measurements taken in the property and the MCS 'corrected' ambient air temperature which accounts for elevation, coastal position and exposure to driving wind.
To obtain an accurate heat pump sizing, the design process is to use the worst-case scenario (generally around -3 deg but this does vary as above) and then the design internal temp (generally 21 deg) and the measurements as detailed above to show the total loss of the property including fabric and ventilation losses.
What's also really important to consider is the heat loss per room as this is what impacts the emitter sizing and therefore the appropriate pipework sizing which then impacts the plantroom layout and by default what heat pumps are suitable.
A proper MCS heat loss will use the actual measurements for your property and a room-by-room assessment to give you and your installer a clear picture of what the heat loss is per room rather than just by property- this seems to be a common error, where the HP is sized correctly but for retrofits the pipework and rads/UFH is not correctly figured to account for the varying loss per m sq of room in the property.
Heat pump consultant and designer at Cirrus Energy. Thinking about installing a heat pump? Or already have one but it’s not performing as it should? Book a one-to-one session with Rob to discuss things here.
The actual values used for MCS are the room design temperature, the insulative values (u-values) for the property as well as the unique measurements taken in the property and the MCS 'corrected' ambient air temperature which accounts for elevation, coastal position and exposure to driving wind.
... and the value for air changes.
Judging by posts on this forum and elsewhere, and comments from a handful of some more independently thinking installers, the air changes are frequently overestimated by a long way. There also appears to be some evidence that the conventional u values for stone walls are way out.
To obtain an accurate heat pump sizing, the design process is to use the worst-case scenario (generally around -3 deg but this does vary as above) and then the design internal temp (generally 21 deg) and the measurements as detailed above to show the total loss of the property including fabric and ventilation losses.
Understood, although I would dispute the word 'accurate'.
Its just an estimate based on a load of assumptions of varying quality and rather dependent on the diligence of the surveyor in taking various things into account. Sometimes it will be good, particularly in more modern houses where the fabric construction is well known and the default assumptions are likely to be fairly close. Sometimes, as in my case, and many other reported cases, it will be a very long way out.
There are installers who have gone on public record on this forum to say that they will not (they say cannot - but its their choice) take into account fabric upgrades which are not visible (for example internal wall insulation) and this can significantly distort the figures. And there are surveyors (or tools?) which count room to room losses but not room to room gains when calculating the house load. As always with computing GIGO applies.
What's also really important to consider is the heat loss per room as this is what impacts the emitter sizing and therefore the appropriate pipework sizing which then impacts the plantroom layout and by default what heat pumps are suitable.
Also understood. The title of this thread is 'Roughly confirming ASHP sizing using existing heat data', not doing the entire design based on existing heat data. The spreadsheet appears to be the way to do this at present, but that doesn't invalidate the concept of at least sense checking (or even calibrating) it against actual measurements. The consequences of specifying a heat pump with far too much capacity can be as serious for the householder as specifying one with too little capacity (its not just about performance!). We should be trying to avoid both!
4kW peak of solar PV since 2011; EV and a 1930s house which has been partially renovated to improve its efficiency. 7kW Vaillant heat pump.
👍
1
ReplyQuote
Share:
Join Us!
Heat Pump Dramas?
Thinking about installing a heat pump but unsure where to start? Already have one but it’s not performing as expected? Or are you locked in a frustrating dispute with an installer or manufacturer? We’re here to help.
