Should I be happy with this decrement delay?

[MikeSharp01]

On Thursday I managed to get the temperature and humidity loggers running in the house and I now have my first scraps of data. The house has no heating running yet so all the heat going in is me working, solar gain and a tiny bit from the logger power supply. I am wondering if the decrement delay is what I should expect, here is the data from midnight last evening to now, (10:46) so about 11 hours.

 

The upper trend is the internal temperature and the lower one the outside temperature derived from the local weather station, about 1 mile from the house. I hope to have my own external one up and running but my PoE scheme out there seems to have a problem. It broadly shows that the temperature started at midnight at about 14.2 and has come down to 13.7 over 11 hours with an outside temperature around 2 ish. Still more analysis to do but should I be happy with that rate of decline? Not sure what i can do if I am not but there are a couple of colder bridges that will be done when the MVHR Inlet and Extracts are fully fitted and there is only plasterboard on about 30% of the walls so far.

 

[SteamyTea]

Assuming your definition of decrement decay is t.K^-1 you have to slice (bin) by the ΔT or bin by the OATs (internal temperature does not have enough variation).

Work out the time between 1K variations, then plot them.

This should produce a natural logarithmic curve i.e. the greater the ΔT, the shorter the time to change.

 

The problem seems to be, when I look at my data, is that it regressed to zero (ideal would be zero i.e. no inputs, no temperature change).

I find it better to look at temperature distribution over a fixed time period, then look at the most likely variation.

 

Maybe when I get home I may look at my data and play with it a bit.

[TerryE]

Mike, I can't recall. Are you using HomeAssistant and do you have Zigbee?  I have a dozen Aqara temperature sensors around the house with a couple stuck on the undersides of external lintels front and back.  I do use the Metoffice day ahead forecast for my weather comp, but the Aqara sensors are easier to use for trending actuals.

 

As to DD, this is driven by the external S.H per m² and U-value.  In our case our DD is well over a day but that is largely down to the ~125mm external stone skin acting as a huge heat capacitor.  This does nothing for the U values but this plus the cellulosic filler gives a long DD.  If we'd had PUR insulation and a render or brick-slip cover then we could still have fantastic U-values but short DD.  Have a look at my ~2016 blog posts on this.  Yes theoretical models, but 9 years later and that how the house behaves in real life.

 

Having a high DD makes using a ToD tariff more effective.

[MikeSharp01]

  On 15/03/2025 at 12:08, TerryE said:

Mike, I can't recall. Are you using HomeAssistant and do you have Zigbee? 

Expand  

No home assistant as yet but will have when the house is finished, we do have a zigbee gateway available and zwave. I will have a look at the old posts. I suppose my concern was how fast it dropped against the delta. We have, will have, fibre cement tile cladding on the outside, 8t, which will add some heat capacity as will the rest of the plasterboard on the inside 4+t. I need to do some more modelling now I have some data to play with along the lines you and @SteamyTea suggest to get a picture of the losses so I can verify potential control schemes - probably very simple ones.

 

[SteamyTea]

  On 15/03/2025 at 12:22, MikeSharp01 said:

need to do some more modelling now I have some data to play with

Expand  

Are you concerned about the tails (extremes)?

 

[MikeSharp01]

  On 15/03/2025 at 12:25, SteamyTea said:

Are you concerned about the tails (extremes)?

Expand  

not clear what you mean?

[SteamyTea]

  On 15/03/2025 at 12:34, MikeSharp01 said:

not clear what you mean?

Expand  

The time it takes for the building to heat, or cool, when there are either very small, or very large, temperature differences?

[MikeSharp01]

Ah- yes I get it. I suppose we should be concerned with just one, in that a delta of 20K is, for this property, probably the worst case likely. To flatten that out will need heat input that matches the losses that manifest themselves in the slope won't it. Therefore the slope must, me thinks, be proportional to the heat losses the question then is wether it does or does not make a difference at the extremes and how to derive energy demand from the slope which can then be used with the delta to work out how to keep ahead of it and maintain the set point temperature. I suspect however that I may be putting the cart before the horse here and somewhat over thinking this. 

[SteamyTea]

  On 15/03/2025 at 13:43, MikeSharp01 said:

I suspect however that I may be putting the cart before the horse here and somewhat over thinking this. 

Expand  

I have been doing that for nearly 2 decades since I did my dissertation on it.

Edited March 15 by SteamyTea

[SteamyTea]

Have just done a very basic calculation on my place, without filtering for inputs.

Heating, on average, takes 15.6 hours per sustained degree increase, cooling, as expected, takes longer at 19.6 hours per degree.

That is based on 6 years data.

 

[MikeSharp01]

  On 15/03/2025 at 14:18, SteamyTea said:

Heating, on average, takes 15.6 hours per sustained degree increase, cooling, as expected, takes longer at 19.6 hours per degree.

That is based on 6 years data.

Expand  

But how does that fit with the delta or does it drop out of the equation somewhere and it must also depend on the rate at which you put energy in or does that drop out as well. I can see that if the inside and the outside are the same temperature and you want the inside 1K above the outside then you need to put energy in to get it there and then sustain it. As I have said here before thermo dynamics was my least favourite subject at Uni and I have avoided anything associated assiduously until now. I guess my lack of attention back then is coming back to bite me in the proverbial bum now trying to get my head around all this. I also suspect that @JohnMo will be chuckling and thinking he should tell me just to KISS away the issue and, after all, perhaps we should just build on the shoulders of giants and not try and re-live all the problems the giants had getting there by starting from first principles ourselves. 

[SteamyTea]

  On 15/03/2025 at 17:13, MikeSharp01 said:

how does that fit with the delta

Expand  

I am looking at that now.

  On 15/03/2025 at 17:13, MikeSharp01 said:

I can see that if the inside and the outside are the same temperature and you want the inside 1K above the outside then you need to put energy in to get it there and then sustain it

Expand  

That as well.

  On 15/03/2025 at 17:13, MikeSharp01 said:

thermo dynamics was my least favourite subject

Expand  

Why, it is basic mathematics, statistics and easy physics.

  On 15/03/2025 at 17:13, MikeSharp01 said:

I also suspect that @JohnMo will be chuckling and thinking he should tell me just to KISS away the issue

Expand  

He is right.

[MikeSharp01]

  On 15/03/2025 at 18:09, SteamyTea said:

Why, it is basic mathematics, statistics and easy physics.

Expand  

I suspect it was the very dry way it was taught, I had done the basic stuff at A level but at Uni it had no applications shown to us unlike the software stuff, mechanical design, electronics & systems all of which seemed to be going places - or so it turned out. I was also somewhat put off by materials science but the good bit of that was, believe it or not, the wood bit - our lecturer had a thing about it. 

[SteamyTea]

Right

Knocked up the basic data, grouped it by ΔT to show the hours it takes to change by 1K.

I have created 4 charts to split the year up, generally the April, May, June, July, August and September have no heating on.  Usually I turn the heating on some time in November and it goes off sometime in March.  The data spans the years 2019 up to 2025, so 6 years.

I will create some charts that show the frequency of the ΔTs, but they will almost certainly be monthly or they will be very busy (LibreOffice Calc is not as fast for chart making as MSO Excel).

 

Note that the y-axis scale changes wildly.

 

 

 

 

 

Generally, as expected, the greater the ΔT, the faster the temperature changes.

I suspect that the anomalies are when the heating is turned on and off, which in itself is useful to know.

[SteamyTea]

Right

Here are the more interesting charts, Decrement Decay, Frequency that they occur, grouped monthly.

As I suspected, the large variations are at the extremes, so don't really matter that much.

Where there are large variations in the middle of the distribution, the decrement decay is still quite large i.e. April taking 15 hours to drop 1K when the mean IAT is 19.1°C.

 

It is also worth remembering that my house is small, terraced and timber framed.  So not an ideal form thermally (it is basically a corridor), and it has disproportionally high window area, at each end.

I also have E7 storage heaters (modified to be only on for up to 4 hours), so a maximum of 16.8 kWh.day^-1, but usually half that.

 

I am not really sure if it is worth looking at hourly data (these charts are based on 6 hourly means so I could split them into heating input times and daylight times, but then when, on average, the decay is several hours, not really important for 1K temperature difference.  My internal temperatures are stable enough, and since the latest improvement (fixed the draughty door and added secondary glazing onto the double glazing), I seem to be using less energy anyway (may have a look as it is only a matter of comparing charts, but tonight).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

[MikeSharp01]

  On 15/03/2025 at 22:21, SteamyTea said:

Right

Here are the more interesting charts, Decrement Decay, Frequency that they occur, grouped monthly.

As I suspected, the large variations are at the extremes, so don't really matter that much.

Where there are large variations in the middle of the distribution, the decrement decay is still quite large i.e. April taking 15 hours to drop 1K when the mean IAT is 19.1°C.

 

It is also worth remembering that my house is small, terraced and timber framed.  So not an ideal form thermally (it is basically a corridor), and it has disproportionally high window area, at each end.

I also have E7 storage heaters (modified to be only on for up to 4 hours), so a maximum of 16.8 kWh.day^-1, but usually half that.

 

I am not really sure if it is worth looking at hourly data (these charts are based on 6 hourly means so I could split them into heating input times and daylight times, but then when, on average, the decay is several hours, not really important for 1K temperature difference.  My internal temperatures are stable enough, and since the latest improvement (fixed the draughty door and added secondary glazing onto the double glazing), I seem to be using less energy anyway (may have a look as it is only a matter of comparing charts, but tonight).

Expand  

That is useful stuff, I will need another glass of wine to get my head around this little lot. Thanks for putting it together.