Posted by: @majordennisbloodnok

↑

Or, as an alternative question, if you have a surface that is kept at a constant temperature, will that surface emit the same amount of heat energy irrespective of what the surface is made of? (e.g. a flat metal surface at 50degC compared with a flat stone surface at 50degC, assuming both surface areas are the same and they are both facing an ambient air temperature of 20degC).

No.  The amount of heat lost/gained by conduction/convection will be dependent on the surface roughness and the nature of the medium in which it is immersed..  The amount of heat radiated depends on the emissivity, for example black bodies surfaces radiate more than polished silver surfaces, as well as the properties of what it is facing.  This is used to good effect in building materials.

Posted by: @majordennisbloodnok

↑

If you have a particular known material (brick, glass, concrete etc.) forming the outside face of a container, you know the ambient temperature outside that container, you know the temperature of the material in question on its external face and you know the surface area of that external face, is it possible just with that information to calculate how much heat that face is transferring to the outside?

No, you need to know the temperature of any material its facing (which determines the radiation loss).  That's why you feel and in fact are hotter when its sunny, even if the air temperature is cool.  You also need to know the properties of the material in which it is immersed, which determines the losses by conduction or convection.

Posted by: @cathoderay

↑

That just leaves the question of whether daily plots, with on/off periods during the day, will work. I can't do the on/off bit as I don't have thermostat controlled on/off data, but I can, with some data manipulation (my fault, because I didn't put all the right variables in all the right places) do daily plots. B

Yes with reservations.  If the house cools materially during the 'off' period then the average IAT will be lower than the 'target' IAT and so the amount of energy lost from the house will be reduced accordingly.  The reduction is actually a lot less than people think.  Many seem to believe that if the heating is on (eg) half the time then the energy input is half as much, which is only true if the house cools instantaneously.  Most houses only cool by a few degrees during several hours of 'off' and so the energy saved is much lower.  The concept of 'average' IAT is a reasonably valid way of working out the correct numbers, because the loss equations are all linear, although its not perfect because the fabric isn't actually a monolithic body, it comprises various components which cool/heat at various rates.

Posted by: @cathoderay

↑

.... hourly plots fail because your system is very volatile.

Hourly plots of OAT vs energy to house may not give a particularly good correlation.  The issue here is that the solid fabric of the house heats/cools much slower than the OAT changes, and the indoor air temperature does something different again based partly on what the inside of the solid fabric does, and partly on incoming ventilation air.  Eventually it all evens out but not within an hour.

I once did an exercise where I had hourly IAT, OAT and energy use for a house of someone on this forum.  The correlation wasn't too bad, but neither was it great.  However by building some of the 'history' into the data  I could get to an r value of about 0.92, a very good correlation both by eye and mathematically.  From the parameters that give the best correlation I suspect its possible to deduce quite a lot about the house fabric including, but not limited to, its heat capacity.

Posted by: @majordennisbloodnok

↑

..Or, as an alternative question, if you have a surface that is kept at a constant temperature, will that surface emit the same amount of heat energy irrespective of what the surface is made of? (e.g. a flat metal surface at 50degC compared with a flat stone surface at 50degC, assuming both surface areas are the same and they are both facing an ambient air temperature of 20degC).

This comes down to the emissivity of the surface. Enamelled steel (radiators) is close to 1 ie: a perfect black body heat emitter.

Chrome towel rails have a very low emissivity by comparison. Compare the output of the same rail in a chrome vs. A powder-coated finish and you'll see the difference.

Posted by: @larry

↑

Posted by: @majordennisbloodnok

↑

..Or, as an alternative question, if you have a surface that is kept at a constant temperature, will that surface emit the same amount of heat energy irrespective of what the surface is made of? (e.g. a flat metal surface at 50degC compared with a flat stone surface at 50degC, assuming both surface areas are the same and they are both facing an ambient air temperature of 20degC).

This comes down to the emissivity of the surface. Enamelled steel (radiators) is close to 1 ie: a perfect black body heat emitter.

Chrome towel rails have a very low emissivity by comparison. Compare the output of the same rail in a chrome vs. A powder-coated finish and you'll see the difference.

Emissivity certainly comes into it but is not strictly the whole story - because radiators aren't predominantly radiators, they are predominantly convectors.  That said, surfaces with high emissivity often have the properties needed for high surface conductivity and thus high heat transfer by convection.

Post removed after reading answers posted whilst this one was being composed.

Posted by: @jamespa

↑

If the house cools materially during the 'off' period then the average IAT will be lower than the 'target' IAT and so the amount of energy lost from the house will be reduced accordingly. 

My suggestion is the heating is always on (on the timer) during any heat loss assessment exercise, set to a constant IAT, the only on/off activity will be from the thermostat, cycling to keep the IAT reasonably constant. It is not an insufferable request if the period is short say a couple of weeks. If the method can be shown to work with timed on/off periods that do cause variations in the IAT, then so be it, but the loss of being in equilibrium introduces all those extra known and unknown variables into play and complexity increases exponentially. 

R value of 0.92? Close only counts in horse shoes and hand grenades! Remarkably my seven day plots posted earlier today with only seven data points are visually very tight, and got an R squared value of 0.9884, implying an R of 0.9942. I need to replicate it before I can believe it! It is all just rather tedious to do, as I have to pull data from different data sets to do the analysis, and they are big data sets with occasional missing data as I said before, so I can't just copy and paste the columns together side by side, I have to manually make the rows line up correctly.

I also have the option of doing a full integration based heating degree day analysis, but if the mean OAT based heating degree days work as well as they appear to, then no need for integrated heating degree days and all the complicated maths that entails. Even just the plain OAT based plot may be good enough (as I've hinted before, for now I see the mean OAT and the HDD based on mean OAT plots as in effect just different ways of plotting the same thing, which is why the slope and R squared values are the same).

Meanwhile there is another more subtle concern I have. All of my different empirical methods of assessing heat loss ultimately end up assessing the same raw data. Could a sort of auto-correlation be at work where the different ways of doing the calculations always end up with the same answer because the starting data was the same? But what if the starting data is wrong? This leads to the question: what is the practical, by which I mean doable with spending a lot of money, gold standard assessment for a heat loss? 

In passing, a thought on MCS: why not get them to cut the heat pump nominal output crap, which makes people think they are getting something they are not (at least not when it matters) and have it a requirement that all heat pump outputs are always given at say 45 degrees LWT, -2 OAT (or thereabouts)? Doing this would also make potential owners more likely to be aware of the importance of performance at lower OATs, simply because the headline output is at a low OAT.   

Posted by: @ianmk13

↑

I'm still running on my gas boiler and have been analysing my half-hourly gas use (and OAT, IAT, flow temp and return temp) for almost two years.

Given this, and the fact you seem to have a fairly good consensus based assessment of your total heat loss, can you do what I did, that is find a period, minimum of a week, preferably longer, when the IAT is reasonably constant, and do a scatter plot of daily energy out (derived from energy in adjusted by your assessment of your boiler efficiency) against daily mean OAT? Then fit the linear regression line, extend it is necessary, and see what the predicted heat loss is at -2 degrees OAT, or whatever your design OAT is?

By the way, if your data is in csv format or can be saved to csv format, there is a very useful cross platform program called q text as data if you haven't come across it that allows you to do sql type queries on csv data. For example this query run on the command line 

q -H -d , "select datetime,amb_24h_mean,htg_24h_kWh_out from midea_24h_data.csv where datetime like '%T00:%' order by datetime desc limit 10" | clip

will pull out the ten most recent midnight rows (my time stamps have T between the date and the time) with the mean ambient (OAT) temp and 24 hour energy delivered variables and copy them to the clipboard so I can just paste them directly wherever I want.    

@cathoderay   I really wouldn't like to hazard a guess at my boiler's efficiency.  Even with flow temperatures around 60C and return below 50C during a DHW session this morning there were still plumes of steam from the flue. I don't know how the boiler controls itself; looking at the circuit schematic I could only see one temperature sensor (on the flow).  Also the burner cycles on/off every 10 minutes - I'm not sure if the boiler itself or the 'smart' 3rd-party timer/thermostat is doing that. Boiler efficiency - or energy input - is a major missing piece of my puzzle

Posted by: @cathoderay

↑

My suggestion is the heating is always on (on the timer) during any heat loss assessment exercise, set to a constant IAT, the only on/off activity will be from the thermostat, cycling to keep the IAT reasonably constant. It is not an insufferable request if the period is short say a couple of weeks. If the method can be shown to work with timed on/off periods that do cause variations in the IAT, then so be it, but the loss of being in equilibrium introduces all those extra known and unknown variables into play and complexity increases exponentially. 

Agreed

Posted by: @cathoderay

↑

Remarkably my seven day plots posted earlier today with only seven data points are visually very tight, and got an R squared value of 0.9884, implying an R of 0.9942.

According to your profile you have an ashp, presumably operating weather compensation.  This being the case I would expect a much better correlation of energy with OAT than would be the case with a fossil burner operating on room temperature, so am not surprised that the correlation is good because that's how the control loop is designed to work.

Posted by: @cathoderay

↑

Meanwhile there is another more subtle concern I have. All of my different empirical methods of assessing heat loss ultimately end up assessing the same raw data. Could a sort of auto-correlation be at work where the different ways of doing the calculations always end up with the same answer because the starting data was the same?

Yes, definitely and I share the concern.

Posted by: @cathoderay

↑

But what if the starting data is wrong? This leads to the question: what is the practical, by which I mean doable with spending a lot of money, gold standard assessment for a heat loss? 

I think we can reasonably assume that utility meter data is correct to the accuracy required for the calculation.  Anything else is suspect,

Posted by: @cathoderay

↑

In passing, a thought on MCS: why not get them to cut the heat pump nominal output crap,

Not really an MCS decision, it's how the manufacturers present the data given that design OAT and FT varies by geography and installation.  They will be guided mostly by EU rules not UK ones (if they differ).

Posted by: @ianmk13

↑

Boiler efficiency - or energy input - is a major missing piece of my puzzle

Energy input (used, consumed) you have, it's what your meter says it is, and you just total it for each day (or subtract previous reading from current if it is a cumulative total). You can then have a stab at estimating efficiency, using @jamespa's guidelines. Or just use 90%. The use whatever efficiency you chose to convert energy input to energy out (provided, delivered). OAT, calculate 24 hour means is you don;t already have them. The just do a scatter plot of 24 hour energy out by mean 24 hour OAT. We are looking for two things, how well they are correlated, and what the trend line shows the energy delivered to be at your design OAT. 

For all the other things, only one matters, the IAT needs to be relatively constant during the assessment (scatter plot) period. How the boiler achieves that doesn't matter, all that matters is it does, and nothing else does - obviously adding in other heaters will do in the results!. A varying IAT on the other hand brings a whole raft of new variables into play, and I haven't yet thought of a way of dealing with that.         

Posted by: @jamespa

↑

According to your profile you have an ashp, presumably operating weather compensation.

@jamespa - you know I do! You can't have forgotten the many discussions we and others have had about it, some of which have got rather warm on occasions. Not so long ago we had a very long debate about whether setbacks saved money, surely you remember that, or that I have set up my own auto-adaption script to adjust my weather comp curve when the actual IAT has moved away from my desired IAT. Etc etc etc.

Posted by: @jamespa

↑

This being the case I would expect a much better correlation of energy with OAT than would be the case with a fossil burner operating on room temperature, so am not surprised that the correlation is good because that's how the control loop is designed to work.

That may or may not be true, but I am not sure it matters. The key point is if the house is in heat balance, ie IAT is relatively constant, then the energy in (delivered) is the energy out (heat loss). How the heat got there doesn't matter, all that matters is you get a good enough estimate of the total quantity. Whether it came from a heat pump, a fossil fuel boiler doesn't matter, what matters is the amount. 

Posted by: @jamespa

↑

I think we can reasonably assume that utility meter data is correct to the accuracy required for the calculation.  Anything else is suspect

This is perhaps a little harsh, and a bit of a presumption of guilt. I prefer a sceptical mindset. That said, your comment doesn't really address the question, which is what is the gold standard assessment for a heat loss? Just using the utility meter energy in isn't going to get up very far... A gold standard test is one that gives you the right (or in rare cases most right possible) answer. In medicine, it might be for example a biopsy, or histology from an operation, and then you look at how other tests eg a blood test or a scan can predict the right answer.

Posted by: @jamespa

↑

Not really an MCS decision, it's how the manufacturers present the data given that design OAT and FT varies by geography and installation. 

MCS publish SCOPs for certified heat pumps on their website, what can't/shouldn't they also publish outputs at low OATs/medium to high FTs? I know the design OAT and FT vary but the point here but about focusing on what matters, a heat pumps real output when it really matters. I think most people living in colder parts of the UK already know they are living in a colder part, and will know, or soon get, that a standard low OAT based on say London may/will not apply to them. But the key thing is, they are now already thinking about their proposed heat pump's output not in cloud cuckoo land, but on a cold day in January.

Product literature currently quotes heat pump output in standard conditions, but those conditions are very lenient. There is nothing to stop having a requirement that the heat pump literature headline output must be at a less forgiving OAT/LWT combination. They need to all be the same, so like is being compared with like.

Posted by: @cathoderay

↑

That may or may not be true, but I am not sure it matters. The key point is if the house is in heat balance, ie IAT is relatively constant, then the energy in (delivered) is the energy out (heat loss). How the heat got there doesn't matter, all that matters is you get a good enough estimate of the total quantity. Whether it came from a heat pump, a fossil fuel boiler doesn't matter, what matters is the amount. 

It matters.

Averaged over a sufficient time period what you say is absolutely true - its the first law of Thermodynamics!

However you were apparently suggesting that a determination could be made from a few points based on the excellent correlation between OAT and energy in your case.  For your case, where the control algorithm responds directly to OAT (because its weather compensated), then you are correct.  However we are discussing how to determine the required heat pump capacity, so the presumption is that what you can measure is an existing fossil burner.  In this case its most unlikely that the control algorithm responds directly to OAT, because weather compensation is not normally implemented for fossil burners in the UK.  Most fossil burners have a control algorithm based on IAT.  Such an algorithm at best lags OAT, and in practice is a slightly complex function of OAT (because of the nature of the fabric) albeit over the long term (say 24-48hrs) follows it.

So with a fossil burner (or more precisely any heating where the control algorithm is linked to IAT not OAT) you can expect good correlation of energy with OAT over a sufficiently long period, but not over a short period such as a few hours.

Posted by: @jamespa

↑

Posted by: @jamespa

↑

I think we can reasonably assume that utility meter data is correct to the accuracy required for the calculation.  Anything else is suspect

 

This is perhaps a little harsh, and a bit of a presumption of guilt. I prefer a sceptical mindset. That said, your comment doesn't really address the question, which is what is the gold standard assessment for a heat loss? Just using the utility meter energy in isn't going to get up very far... A gold standard test is one that gives you the right (or in rare cases most right possible) answer. In medicine, it might be for example a biopsy, or histology from an operation, and then you look at how other tests eg a blood test or a scan can predict the right answer.

By 'suspect' I mean may possibly be insufficiently accurate and thus needs to be intelligently questioned.  I would hesitate to say one should presume guilt, but at the same time neither would I presume innocence.  It does of course depend on the accuracy one is seeking. 

In response to the 'gold standard' question, I am not sure there is a reasonably accessible (and certainly there isn't a defined) 'gold standard'.  To create such a standard one would have first to specify solar gain, wind speed and direction and doubtless several other factors which affect the loss.  One would also have to specify the probability of the occurrence of the specific weather conditions and whether that was based on past climate of some future climate.  In the case of the latter one would have to specify the assumptions about the future climate.

All of this is possible but, I would suggest, well beyond the capabilities of the heating industry and beyond what is reasonably sensible.

There are a couple of warehouses (labs) where real houses have been built in a controlled environment.  This is essentially the 'gold' standard, but is not practically accessible in cases of interest.

I would suggest that, rather than looking for a gold standard, one needs to look for a method (or combination of methods) which in most cases gives an answer to an accuracy sufficient that any material design decision is not misled by any inaccuracy.  That's a bit case dependent, but order 10-20% is possibly around the right ballpark except for (low loss) houses where the decision is dominated by the available minimum design capacity of available heat pumps and DHW requirements, not by the house loss and so the tolerance on heat loss determination is much greater. 

Posted by: @jamespa

↑

Posted by: @jamespa

↑

Not really an MCS decision, it's how the manufacturers present the data given that design OAT and FT varies by geography and installation. 

 

 

MCS publish SCOPs for certified heat pumps on their website, what can't/shouldn't they also publish outputs at low OATs/medium to high FTs? I know the design OAT and FT vary but the point here but about focusing on what matters, a heat pumps real output when it really matters. I think most people living in colder parts of the UK already know they are living in a colder part, and will know, or soon get, that a standard low OAT based on say London may/will not apply to them. But the key thing is, they are now already thinking about their proposed heat pump's output not in cloud cuckoo land, but on a cold day in January.

Product literature currently quotes heat pump output in standard conditions, but those conditions are very lenient. There is nothing to stop having a requirement that the heat pump literature headline output must be at a less forgiving OAT/LWT combination. They need to all be the same, so like is being compared with like.

I would agree that publishing actual outputs at a greater range of conditions than are required by the BS/EN standards (basically A2/W35, A2/W55, A7W35 and A7W55 so far as I can tell, plus SCOPs at specified conditions) would be a major step forward.  

For me its the standards that need changing here not MCS, and MCS should be campaigning for a change to these (if it isn't already).  I don't see why MCS would want to risk (or can reasonably be compelled to risk) publishing data that the manufacturers are not compelled to publish, but I can see why it would (and should) campaign for greater transparency from the manufacturers.   The same, incidentally, holds for noise.