@larry - I have now read the SPAB U value report in full, and it makes interesting reading, despite in places getting tangled up in putting up a basecase for problematising the English language eg "The calculation of U-values for traditional walls is further problematised by the lack of basecase thermal conductivity data for most UK vernacular building materials". Once we have got past this architectural vernacular, the findings strongly suggest that standard calculated heat loss assessments significantly over-estimate the heat loss from traditional buildings, in particular those with solid stone walls:
"A further average in situ U-value of 1.36 W/m^2K was calculated solely for the 32 solid stone or brick walls in the study or 1.42 W/m^2K for the 21 stone walls and 1.24 W/m^2K for the 11 solid brick walls. These figures are lower than the U-values given for stone (2.4 & 2.1 W/m^2K) and brick walls (2.1 W/m^2K) in Table S6: Wall U- values for England and Wales in Appendix S of the Standard Assessment Procedure (SAP) 2009 document used in the calculation of SAP ratings for existing buildings (RdSAP)."
These differences are not trivial. For solid stone walls, which is what I have, the SAP rating is 2.4 & 2.1 W/m^2K, whereas the measured average was 1.42 W/m^2K, meaning the SAP figures over-estimate the measured values by around 60%.
@jamespa - this may be the until now missing explanation for the heat loss discrepancy (~12kW calculated vs ~8kw as measured by heat delivered) for my building. The Freedom spreadsheet calculations used to calculate my heat loss used mostly stone 300mm (U value 2.75) and stone 450mm (U value 2.23), and came up with a total heat loss of 12.3kW. If this is an over-estimate of my actual heat loss by say 60%, then the actual heat loss is ~7.7kW (at design outside air temp -1.8), which just happens to be my whole house measured heat loss, based on the actual heat delivered to keep the house at the design indoor air temp, based on this chart (posted again here for ease of reference):
The chart plots, for January of this year, the energy delivered for each hour (value in kWh, so we can also say for those challenged by energy vs power that the average power delivered over that hour is the same value in kW) for each average outside air temp for that hour, with the inside air temp at or around the desired 18 degrees. The energy out (delivered) is calculated from the heat pump's flow rate and LWT/RWT delta t and the specific heat of the circulating fluid, and insofar as anything can be measured on a heat pump, is the actual output. Using the trend line (standard linear regression), the heat loss at -1.8 outside is give or take 7.7kW.
@larry - I have now read the SPAB U value report in full, and it makes interesting reading, despite in places getting tangled up in putting up a basecase for ploblematising the English language eg "The calculation of U-values for traditional walls is further problematised by the lack of basecase thermal conductivity data for most UK vernacular building materials". Once we have got past this architectural vernacular, the findings strongly suggest that standard calculated heat loss assessments significantly over-estimate the heat loss from traditional buildings, in particular those with solid stone walls:
"A further average in situ U-value of 1.36 W/m^2K was calculated solely for the 32 solid stone or brick walls in the study or 1.42 W/m^2K for the 21 stone walls and 1.24 W/m^2K for the 11 solid brick walls. These figures are lower than the U-values given for stone (2.4 & 2.1 W/m^2K) and brick walls (2.1 W/m^2K) in Table S6: Wall U- values for England and Wales in Appendix S of the Standard Assessment Procedure (SAP) 2009 document used in the calculation of SAP ratings for existing buildings (RdSAP)."
These differences are not trivial. For solid stone walls, which is what I have, the SAP rating is 2.4 & 2.1 W/m^2K, whereas the measured average was 1.42 W/m^2K, meaning the SAP figures over-estimate the measured values by around 60%.
@jamespa - this may be the until now missing explanation for the heat loss discrepancy (~12kW calculated vs ~8kw as measured by heat delivered) for my building. The Freedom spreadsheet calculations used to calculate my heat loss used mostly stone 300mm (U value 2.75) and stone 450mm (U value 2.23), and came up with a total heat loss of 12.3kW. If this is an over-estimate of my actual heat loss by say 60%, then the actual heat loss is ~7.7kW (at design outside air temp -1.8), which just happens to be my whole house measured heat loss, based on the actual heat delivered to keep the house at the design indoor air temp, based on this chart (posted again here for ease of reference):
The chart plots, for January of this year, the energy delivered for each hour (value in kWh, so we can also say for those challenged by energy vs power that the average power delivered over that hour is the same value in kW) for each average outside air temp for that hour, with the inside air temp at or around the desired 18 degrees. The energy out (delivered) is calculated from the heat pump's flow rate and LWT/RWT delta t and the specific heat of the circulating fluid, and insofar as anything can be measured on a heat pump, is the actual output. Using the trend line (standard linear regression), the heat loss at -1.8 outside is give or take 7.7kW.
Interesting
Yet more evidence for the proposition that reliance solely, or even primarily, on the fabric spreadsheet method is questionable.
I can't see the industry view changing anytime soon though. The lengths people go to even on this forum to defend it defy logical explanation, and it's built in complexity and relationship to BS whatever is a cast iron guarantee that the installer can't be challenged and is thus protected. Graham Hendra seems prepared to break ranks (listen to his comments three minutes into the video on ashp sizing on this forum) and there are clearly a few other individuals who will think out of the box, but not make a point of speaking about it, however the vast majority stick to the mcs ordained script.
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.
I'm looking to establish the best way to assess heat loss, that is MCS compliant.
Can the results of blower tests be factored in to the air changes?
I've never had one done but I assume if I get a whole house result then I can factor it in, on a room by room basis somehow.
Im late to reply because we have had visitors but...
That is two different questions:
1. What is the best way to assess heat loss?
2. What is MCS compliant
In practice there is a third question namely what will an installer accept!
I have long argued that the fabric spreadsheet is a case of GIGO, and there appears to be far too much evidence accumulating that it is designed principally for the protection of installers rather than householders.
My own experience, documented on the video on this forum about ASHP sizing, is probably extreme, but illustrates the problem. In short I has 2 full 3 hour surveys both of which came out with an estimate of 16kW. I paid for one of these (the other was free - not sure why) so I requested the full calculations. They had ignored invisible fabric upgrades which I had made a big point of telling them about, and had included room to room losses (but not room to room gains) in the whose-house calculation. The argument which was made for ignoring invisible fabric upgrades is that they 'cannot be verified'. I can only presume that the survey I got for free had made similar basic errors.
If you correct for these obvious mistakes, but otherwise use the MCS assumptions you get to 10.5-11kW
The measured loss, determined over two years using a variety of analyses of smart meter readings, is 7.5kW, and if I cap my boiler output at 8.5kW (which is the lowest it will go) I am still warm.
The difference between 7.5kW and 16kW is rather significant to say the least in terms of heat pump physical size, the planning implications, the implications for radiator sizing and the implications for upgrades of pipework. Its far too big to ignore. However, whenever I suggest on this forum or others that experimental measurements should somehow be factored into the assessment of heat loss if only as a sense check, I get a barrage of attacks that they have many unknowns/uncertainties. That is of course the case, but it is also the case for the GIGO spreadsheet method. Who in their right mind ignores experimental results, even if their are uncertainties, in favour solely of theory based on frequently extremely uncertain data?
The plain fact of the matter that the GIGO spreadsheet protects the installer, by giving them a rigid method to follow which is more or less guaranteed to overestimate the loss and therefore wont result in a cold customer, albeit that it might result in the customer having to spend far more money, endure far more disruption, and have far more difficulty with planning than is necessary. Any variation on this rigidity introduces ambiguity or discretion behind which the installer cannot hide.
So to answer the question 'What is the best way to assess heat loss', I would say using as much complementary data as you can get. Do the spreadsheet calculation for sure, and if you get someone else to do it check their workings. If you heat the house for a reasonable part of the time and have consumption data, work out what that tells you. If you have a measurement of air changes, factor that in. And if the various methods don't agree, try to work out why. Then, if the uncertainty/difference between 'experiment' and 'theory' matters makes a material difference (for some houses it wont) find an installer who will take into account data other than the GIGO spreadsheet and reject those who will not.
What I would not do, in a retrofit where the fabric has been subject to multiple 'adaptions' at various times and to various standards, is rely solely on the GIGO spreadsheet as completed by a MCS surveyor! The British Standard on which the MCS GIGO spreadsheet is based was, so far as I can ascertain, designed for new build, where the fabric is both simple and relatively well known. The majority of the heat pump market that we need to tackle is retrofit however, often retrofits that span several iterations of building standards. We need, IMHO, to get a lot more intelligent in how we use available data.
This post was modified 5 months ago 6 times by JamesPa
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.
Mvhr systems can be included in msc calculations since ours was by 3 potential installers. Two had to go away and ask their boss and one needed help from the helpline of the software they used.
The inputs needed are flow rate on the Mvhr and efficiency (nominal). There are tweaks on software since (if I understand and recall a discussion correctly) the option only shows on a more modern house, and you may have to pretend the house is newer.
Background air leakage from a door test divided by 20 (nominal, very dependent on wind speed and how exposed the house is).
2kW + Growatt & 4kW +Sunnyboy PV on south-facing roof Solar thermal. 9.5kWh Givenergy battery with AC3. MVHR. Vaillant 7kW ASHP (very pleased with it) open system operating on WC
I've been looking into the blower ACH50 / 20 method according to the SAP guidance. I’ve also just read a study that indicates this divisor could actually be as high for 37 in the UK.
This all ties in with the fact that your associate in the video has never seen a blower test that indicated over 1 ACPH.
I believe we're permitted to take further evidence into account within the MCS process, so surely a blower test should me mandatory, at least for pre-2000 builds.
Would it be considered reasonable to adjust the notional room by room ACPH on a volumetric basis, to give a whole house ACPH equivalent to the blower test.
First “no blower test above 1ACH”. If these are done on a building site on a typical design, once this passes the rest of the houses are built the same. From discussion with the blower test engineer and tradesmen the true process is the trades take great care to not leave gaps and miss off tape on the test house. The test engineer tests it and if it fails they go round find the leaks fix them with tape or necessary method. Once the test is less than 1ACH the whole estate with that design is passed. Anyone care to guess how carefully the other houses will be built?
So there are no test certificates with more than 1ACH but there are lots of houses with higher leak rates.
For our house I used fairly arbitrary values for the ACH in each room in Heat punk, (they were all so small as to be ignored.) But if the MVHR has been set up correctly there are ventilation rates for each room according to use in the building regs, bathrooms, kitchens etc. The building inspector didn’t ask to see our set-up figures although we did have the measurements for each room.
2kW + Growatt & 4kW +Sunnyboy PV on south-facing roof Solar thermal. 9.5kWh Givenergy battery with AC3. MVHR. Vaillant 7kW ASHP (very pleased with it) open system operating on WC
very dependent on wind speed and how exposed the house is
I have noticed that as well as solar gain, my house also gets wind chill on wet windy days, especially if the wind is in the wrong quarter. Yet, so far as i know, none of the heat loss calculators I have seen take these factors into account.
I also noticed earlier when looking at my heat loss calculations that the biggest single contributor to heat loss in the calculator was wall heat loss, at 58% of the total. Just correcting that (for example as I did above, by putting in more realistic U values) had a very significant effect on total heat loss.
Midea 14kW (for now...) ASHP heating both building and DHW
We get extra heat loss on wet and windy days too, even the best heat loss model, the passive house one (which does allow for solar gain and shading) doesn’t cover that. But any physicist or person who has used one of those ancient techniques of a piece of muslin cloth in a bowl of water to keep milk slightly cooler knows that it is effective in reducing temperature. The thought experiment is how much more heat is removed by putting a fan next to the milk bottle compared with no forced air movement.
2kW + Growatt & 4kW +Sunnyboy PV on south-facing roof Solar thermal. 9.5kWh Givenergy battery with AC3. MVHR. Vaillant 7kW ASHP (very pleased with it) open system operating on WC
We get extra heat loss on wet and windy days too, even the best heat loss model, the passive house one (which does allow for solar gain and shading) doesn’t cover that. But any physicist or person who has used one of those ancient techniques of a piece of muslin cloth in a bowl of water to keep milk slightly cooler knows that it is effective in reducing temperature. The thought experiment is how much more heat is removed by putting a fan next to the milk bottle compared with no forced air movement.
In fairness it's unlikely that any model will ever take all the variables into account, particularly since the variables may occur in multiple combinations with varying (possibly unknow) frequencies.
That's another reason I advocate an element of experimental measurement over a period of time, to complement the fabric based calculation.
Two or three years worth of data, if you have them, tell you a lot about what happens in practice. Of course there may still be events falling outside these parameters, but they will be rare, at least until the gulf stream shuts down at which point none of the models are valid. Auxiliary heating or a jumper can deal with rare events
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.
In fairness it's unlikely that any model will ever take all the variables into account, particularly since the variables may occur in multiple combinations with varying (possibly unknow) frequencies.
In fairness, this is what I have been saying for some considerable time. It is why whatiffery fails. Using crass numbers eg stone wall U values doesn't help either.
That's another reason I advocate an element of experimental measurement over a period of time, to complement the fabric based calculation.
This is perhaps where we differ. Instead of having an element of experimental data complement the model, I fit the model to the data. It is a question of primacy, which comes first. Here we go: full data from my system for the last week:
There are some haywire COP values when the OAT goes above 15 degrees, which just happens to be the right hand end point on my weather compensation curve - perhaps the heat pump puts a cork in it when the OAT gets to these temps - but the important thing to note is the IAT is pretty stable, ie the house is in equilibrium, the heat energy being put in matches the heat being lost, and IAT stays the same. Since we believe there is an underlying linear relationship between the OAT and the heat loss, we can plot one against the other, hourly values, and do the regression (note this is space heating only, I have excluded the DHW hours):
As we can see, the two are tightly correlated. The remaining variation ('the residuals') is caused by all those factors the model based approaches don't know about, solar gain, wind chill, Grandad over-exerting himself, the mother of all Sunday lunches, someone left the door open etc. But that's OK, because it turns out that they they largely cancel themselves out. Here are the residuals (what we want to see, and it is what we see, is that they are fairly evenly spread about the zero line, suggesting there is no hidden bias at work, and that doing a regression on the data is a reasonable thing to do):
That really is as far as I should go, but I have taken the liberty of extrapolating the regression line in the regression plot back to -2 OAT, to see what the predicted (on the basis of this sample from a relatively mild week in October) heat loss per hour is at that design OAT and bingo, at about 7.6kWh, it is very close to the value from the January sample for this year that I posted earlier (7.7kWh). This small difference is well within measurement error, and frankly is too small to worry about - unlike the difference between these values and the whatiffery based MCS/Freedom spreadsheet calculator based value of 12.3kW(h). Two experimental samples from different times in the year, same result, starts to look fairly convincing.
I am lucky enough to have hourly data, which in turn is based on automatic monitoring of minute by minute data, and can do these plots to determine my heat loss empirically. At some point, I might do the same analysis, but use daily data (which it is practical to record manually, meaning anyone can do it, for any heating system), and see how that works out.
Midea 14kW (for now...) ASHP heating both building and DHW
In fairness it's unlikely that any model will ever take all the variables into account, particularly since the variables may occur in multiple combinations with varying (possibly unknow) frequencies.
In fairness, this is what I have been saying for some considerable time. It is why whatiffery fails. Using crass numbers eg stone wall U values doesn't help either.
In fairness, @jamespa hasn't said whatiffery fails. He went on to say...
...I advocate an element of experimental measurement over a period of time, to complement the fabric based calculation.
I'm not hearing many people saying that heat loss calculations should be used exclusively. However, they do give at least a starting point and that's not a failure. If someone needs to plan an installation, the ballpark figure is a much better starting point than no figure at all.
What @jamespa is saying and what many have said repeatedly on many threads is that combining that ballpark with empirical data where it is possible to obtain will give a significant improvement on the reliability of the estimate to work around. That's not the same as saying "whatiffery fails".
My concern is that in highlighting the limitations of the heat loss calculation process we are inadvertently lending weight to the idea of system design with no heat loss estimate at all; relying, perhaps, on the installer's intuition and/or experience. And there be monsters in that territory...
105 m2 bungalow in South East England
Mitsubishi Ecodan 8.5 kW air source heat pump
18 x 360W solar panels
1 x 6 kW GroWatt battery and SPH5000 inverter
1 x Myenergi Zappi
1 x VW ID3
Raised beds for home-grown veg and chickens for eggs
That's another reason I advocate an element of experimental measurement over a period of time, to complement the fabric based calculation.
This is perhaps where we differ. Instead of having an element of experimental data complement the model, I fit the model to the data. It is a question of primacy, which comes first. Here we go: full data from my system for the last week:
I don't think we differ. I would also do the same on the basis that (good) experiment trumps theory. I was being a bit 'conservative' in my statement and also conscious that there are legitimate examples where the experiment isn't very useful or at best difficult to interpret, for example where heating is very 'part time '. However where there is good data for a house that is heated a good proportion of the time (more precisely that is warm to within a few degrees of he 'design' OAT for most of the time) I would argue that experimental data is superior to model in most if not all cases.
I'm not hearing many people saying that heat loss calculations should be used exclusively.
The majority of the installers I have encountered in my 2 year journey seem to take the view that the fabric spreadsheet method is definitive so far as I can tell, and some on this forum do also.
My concern is that in highlighting the limitations of the heat loss calculation process we are inadvertently lending weight to the idea of system design with no heat loss estimate at all; relying, perhaps, on the installer's intuition and/or experience. And there be monsters in that territory...
Fair comment, although we are also advocating that specific additional data be brought to bear, not just saying that the current method is pants!
That said my strong suspicion is that, with a multi-dimensional analysis of a year's worth of half hourly data, more sophisticated than is possible with a spreadsheet, it would be possible to come up with a very accurate estimate and also interesting information about the effects of solar gain, wind etc. There is a phd or two in this I would think, and a very useful one at that.
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.
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.
