System sizing based on actual consumption - a better way?

[JamesPa]

17 minutes ago, Carrerahill said:

I won't read the whole thread, out of laziness, but, the way to size heating systems properly is to do heat loss calcs for the building, then size the emitters (usually with 10% margin) then size the source.

 

I have a spreadsheet which you key all the figures into and it spits out the heat loss, you key in your U values for walls, glazing, floor & ceiling, there is a column for infiltration and that's you. 

 

We are doing some 3 bed flats at the moment and the heat loss for the whole flat is 4.9kW. Therefore the ideal source would be about 5kW, yet in plenty of flats like this you will find a 20kW boiler (I understand the water generation argument there) but most sources are oversized. 

 

To save you reading the whole thread, my 'challenge' quoted below (24/05/2023 at 15:59) is what started it off.

 

Whilst I wouldn't argue with what you say in a straightforward house which is relatively new, un-extended and where the fabric has not been the subject of piecemeal upgrades, I think the problem with your blanket statement is that, in a typical retrofit, there can be lots of unknowns.  The infiltration loss is a complete guess and u values of upgraded components are not always known, particularly if they are not visible (eg wall insulation).  As a real example my 1930s solid brick house gets a rating by 2 MCS surveyors of 16kW (each after a survey lasting 3 hours).  if the calculations are done with the best estimate I have of U-Values, I get to 11kW.  The actual measured loss is 7.5kW.  The two MCS surveyors ignored the invisible fabric upgrades (I made a point of telling them about them more than once) and I reckon that the reason my calculations still overestimate the load is infiltration loss (I used the MCS assumptions).

 

I hear a similar story far too often and find it difficult not to conclude that the spreadsheets are pseudo-science in many cases (most of our housing stock is older rather than new, and most older houses have had upgrades of some sort done to them so quite likely in most cases).  They look impressive but GIGO applies.  I have a very strong feeling that the conventional wisdom is fatally flawed in many cases of interest.

 

 

On 24/05/2023 at 15:59, JamesPa said:

From my reading on here, listening to Heat Geek videos, personal experience with prospective installers, and comments and evidence from another forum Im becoming increasingly convinced that poor system sizing (generally oversizing) is one of the key factors to root out over-engineering leading to poor performance, significant excess cost and significant unnecessary disruption. 

 

The fact is that, in a typical retrofit, there are real unknowns due to:

 

• invisible fabric upgrades (eg wall insulation)
• the unknown performance of double glazing, which can have a U value anywhere between 2.8 and 1.4 
• later extensions of a different construction to the original building
• the unknown ventilation rates

 

It is thus perhaps inevitable that theoretical calculations err towards oversizing, in my case and others of which I have heard, by a factor of two. 

 

However the destructive effects of this are quite devastating, in terms of cost, disruption and the negative effects on performance.  Its all very fine blaming installers/surveyors (as I have done in spades because they have ignored what I have told them about the fabric), but, realistically, when they cant verify the key parameters of the key calculation, what else are they going to do if they are anxious to avoid 'im too cold' service calls?

 

The more I consider this problem the more convinced I become that this cannot continue and at the same time lead to satisfactory outcomes.  Heat pumps with a high modulation depth help, but don't fix the consequent overengineering of the system.  Range-rated heat pumps would similarly help, but again don't fix the consequent overengineering of the system. 

 

Of course there is an alternative, namely experimental measurements of whole house heat loss.  With a smart meter, half hourly meter readings are available and can readily be correlated with average OAT/degree days to get the required load (and with a bit more manipulation the 'thermal mass', which could help with setting up control parameters).  If no smart meter is present then annual consumption plus some information about heating patterns could, at worst, be used as a 'sense check' on the calculated figures.  In another thread there are references to companies that make special measurements to determine experimentally the heat loss which are a further alternative

 

Of course some room by room calculations will be needed to size radiators, but getting a radiator too big is nothing like as serious a problem as getting the system too big.

 

Discuss!  Has anyone got a better idea of how we might practically deal with the real fabric uncertainties in a typical retrofit?  If not then how else do you suggest we square off the problem that we simply do not, in many cases, know, with any degree of accuracy, the key parameters to do the theoretical calculation that is the basis of almost all much heat pump system design.

 

 

 

Edited May 31, 2023 by JamesPa

[Carrerahill]

3 minutes ago, JamesPa said:

To save you reading the whole thread, my 'challenge' quoted below (24/05/2023 at 15:59) is what started it off.

 

Whilst I wouldn't argue with what you say in a straightforward house which is relatively new, un-extended and where the fabric has not been the subject of piecemeal upgrades, I think the problem with your blanket statement is that, in a typical retrofit, there can be lots of unknowns.  The infiltration loss is a complete guess and u values of upgraded components are not always known, particularly if they are not visible (eg wall insulation).  As a real example my 1930s solid brick house gets a rating by 2 MCS surveyors of 16kW (each after a survey lasting 3 hours).  if the calculations are done with the best estimate I have of U-Values, I get to 11kW.  The actual measured loss is 7.5kW.  The two MCS surveyors ignored the invisible fabric upgrades and I reckon that the reason my calculations still overestimate the load is infiltration loss (I used the MCS assumptions).

 

I hear a similar story far too often and find it difficult not to conclude that the spreadsheets are pseudo-science in many cases (most of our housing stock is older rather than new, and most older houses have had upgrades of some sort done to them so quite likely in most cases).  They look impressive but GIGO applies.  I have a very strong feeling that the conventional wisdom is fatally flawed in many cases of interest.

 

 

 

Agree flawed for retrofit - less so for new build and I hope it works as that is how heating is sized! 

[JamesPa]

15 minutes ago, Carrerahill said:

Agree flawed for retrofit - less so for new build and I hope it works as that is how heating is sized! 

Good, agreement.  It doesn't work for retrofit in a fair number of cases, so far as I am gathering, so it needs to be done differently.  Retrofit is the majority market.  We build 200k new houses each year and retrofit 1.4M gas boilers, which needs to become 1.4M heat pumps.  Sizing them wrongly is a disaster.

Edited May 31, 2023 by JamesPa

[sharpener]

 

 

1 hour ago, JamesPa said:

Unfortunately it's a thermal insulator as well as an absorber of sound.

 

17 hours ago, sharpener said:

leaving staggered ventilation openings with no line of sight through

[JamesPa]

9 minutes ago, sharpener said:

 

 

1 hour ago, JamesPa said:

Unfortunately it's a thermal insulator as well as an absorber of sound.

 

17 hours ago, sharpener said:

leaving staggered ventilation openings with no line of sight through

Yes, and if I had built the boat myself knowing what I know now, that's probably what I would have done.  That said, space is at a premium for accommodating anything staggering (other than the skipper and crew when returning from the pub, that is.)

[sharpener]

1 hour ago, JamesPa said:

Yes, and if I had built the boat myself knowing what I know now, that's probably what I would have done.  That said, space is at a premium for accommodating anything staggering (other than the skipper and crew when returning from the pub, that is.)

 

Sorry, didn't realise you were talking about the boat. We used to have a Rival 32 with a very noisy 2 cyl Volvo engine, under the cockpit there was some fancy foam insulation faced with a perforated white top sheet, pretty haphazardly installed and so not very effective. Soundproofing is as much an art as a science.

[JamesPa]

2 hours ago, sharpener said:

 

Sorry, didn't realise you were talking about the boat. We used to have a Rival 32 with a very noisy 2 cyl Volvo engine, under the cockpit there was some fancy foam insulation faced with a perforated white top sheet, pretty haphazardly installed and so not very effective. Soundproofing is as much an art as a science.

No apology needed, a welcome opportunity for some levity amongst all the serious stuff!

 

As regards the serious stuff, I don't think anyone has seriously disagreed with the challenge with which I initiated this thread. 

 

I am awaiting a response from MCS about the flawed nature of their procedures for the majority use case, ie retrofits of properties that have been 'messed around with' (Heat Punk declined to comment).

 

The question is why am I (apparently) the only person talking about it, albeit that lots are talking about the industry failings.  Surely this is exactly the sort of thing that the BUS should be stimulating the industry to sort out and, if it isn't achieving that, then what's the point?

Edited May 31, 2023 by JamesPa

[DanDee]

29 minutes ago, JamesPa said:

The question is why am I (apparently) the only person talking about it, albeit that lots are talking about the industry failings.  Surely this is exactly the sort of thing that the BUS should be stimulating the industry to sort out and, if it isn't achieving that, then what's the point?

 

It's all about public funds money making. Democracy comes with many edges, more than a sword.

[sharpener]

On 25/05/2023 at 16:43, billt said:

I think you need to look at it more closely.

 

Our house is a hotchpotch of additions carried out at different times with different construction methods and it is possible to model that in the heat punk calculator.

 

You can enter fabric losses for each element in each room separately as well as air changes for each room. It should come up with a pretty accurate heat loss figure.

 

In my case, very closely indeed(!)

 

I have now done a more comprehensive heatpunk model including internal walls and individual rooms, which showed that for some reason the previous shell model was only half my house [it is two barns end-to-end]. Now it has been fixed, the total MCS heat requirement is a shade under 15kW. Since the Vaillant A+ 12 is very generously specced (unlike the Samsung HT 14 which is positively mendacious) I think that will suffice. There is MVHR and additional heating in the form of an AGA and a WBS which of course MCS won't allow for.

 

Result is remarkably close to the central figure from the Heat Geek cheat sheet, 75 W/m^2. And reasonably consistent with the parsimonious 6kW average oil consumption required to heat only the zones I am using at any one time (kitchen plus two bedrooms or the living room).

 

Still nothing heard from the two installers who came a week ago. A third contacted me courtesy of the Vaillant "find an installer" scheme, Daikin seem to have given up completely.

[JamesPa]

26 minutes ago, sharpener said:

 

I have now done a more comprehensive heatpunk model including internal walls and individual rooms, which showed that for some reason the previous shell model was only half my house [it is two barns end-to-end]. Now it has been fixed, the total MCS heat requirement is a shade under 15kW. Since the Vaillant A+ 12 is very generously specced (unlike the Samsung HT 14 which is positively mendacious) I think that will suffice. There is MVHR and additional heating in the form of an AGA and a WBS which of course MCS won't allow for.

I don't know if the vaillant 12kW stretches to 15 but you might need to lie a bit about the construction if it doesn't.  MCS as you know requires that the system heat the whole house unless not reasonably practical so if the MCS calculation says 15 they might feel obliged to fit 15+.  Obviously this is very installer dependent, hopefully you find a reasonable one.   I'm sure you have worked this out yourself by now anyway.

[sharpener]

Yes, it will. This is what I meant by generously specced (columns are 35/40/45/50/55C flow), the superiority of the R290 is obvious cf the Samsung R32 offering:

 

 

One issue is the design OAT, the heatpunk s/w gives -1.3 for my postcode whereas the MCS table has -0.2. Another is making sure they do not apply any uplift for near a river (should be an advantage IMO), or intermittency. Fortunately the EPC surveyor was generous in assessing the loft insulation thickness to avoid disqualification for BUS. (Makes damn all difference to the overall picture, payback for adding more is ~10years if DIY, awful job, ?twice that if contracted).

 

Have just spotted I can reduce the U value of the walls from 2.27 to 1.86 by selecting the stock "610mm solid stone" instead of 600mm "sandstone". But it only reduces the heat loss by a few 100 W, further investigation of this required methinks!

[sharpener]

2 hours ago, sharpener said:

But it only reduces the heat loss by a few 100 W, further investigation of this required methinks!

 

So I find the Heatpunk software has some worrying inconsistencies. As well as the swapped rad selection boxes noted upthread, it seems that going back and changing the wall materials gives unreproducible results, if I reverse the changes I get an answer about half of what I started with.

 

To get a true comparison between the stock "610mm solid stone" and the custom "sandstone 600mm" I needed to start again and build a new model identical save for the wall materials - which is annoyingly time-consuming as there does not appear to be a "copy floorplan" feature.

 

Result is that the "610 mm solid stone" (U=1.86) needs 12180W whereas the "sandstone 600mm" (U=2.27) needs 13776W.

 

So it is clear which one the surveyor needs to use to get the "right" answer!