Help calculating U-Value of ground floor...

[JohnW]

Hi

I think I understand how U-Values are calculated e.g.

1. (Layer thickness) X Lambda = R in m² K/W 
2. Add up all the R-values for each of the layers and
3. 1/(Total R-Value) = U-Value in W/m² K

 

However if you look at the calculations from my "As-designed" SAP 2009 report below, the calculated Total R-Value = 5.844 m² K/W and if you take the inverse of that the U-Value is 1/5.844 = 0.171 W/m² K and not the 0.14 shown on the report below. 

 

Can anyone please explain how the 0.14 was achieved...what am I missing???

 

Thanks

[PeterW]

You need to use the P/A ratio which is the perimeter and area

 

I expect the worksheet does that behind the scenes. 

[Moonshine]

i think that you need to account for the er

Just now, PeterW said:

You need to use the P/A ratio which is the perimeter and area

 

I expect the worksheet does that behind the scenes. 

 

how is it applied to 0.17 to get 0.14?

[Ed Davies]

The point is that the earth under the floor will provide some resistance to be added to that provided by the materials you lay on top. How much is very dependent on the properties of the soil (dry sand will insulate a lot better than wet mud) and also on the geometry of the house because heat loss has to go sideways. A round house (or, more likely, square one) will lose a lot less through the soil than a long thin one. This could all be calculated using finite-element analysis or whatever but in practice is it's approximated based on the perimeter to area ratio, as @PeterW says. Sorry, don't know the details of the calculation.

 

But, I'm not sure that chart you quote is quite right. It gives “Ext surface” R = 0.040. That's the value for an external surface to air, I don't think it applies for an interface to mud and I think the value should be approximately zero. However, it's a tiny fraction of the total resistance anyway so will not make any significant difference.

[A_L]

1 hour ago, JohnW said:

Can anyone please explain how the 0.14 was achieved...what am I missing???

 

The reason is that the thickness of one of the layers, the soil, varies across the area of the floor. The heatloss path is floor surface - soil - external air. Therefore at the edge of the slab there is little/no soil and the local R value is approximately the higher one. The area at the middle of the floor has the thermal resistance of a considerable thickness of soil and thus a higher R-value. The 0.14 U value is the weighted average of the whole floor. If you try using floors of the same area but different P/A ratios you will see the effect, since higher P/A ratios imply less 'square' shape which have greater exposed perimeters for a given area,  e.g. the common perimeter of two adjoining semi's would not be included.

[Jeremy Harris]

@Ed Davies is right, to get a true figure the interface thermal resistance needs to be accounted for, although for a well-designed slab with an insulated perimeter it's highly variable, depending on ground conditions, and in practice it doesn't usually have a significant impact.  I did model it for our slab and found I had wasted my time, as the edge loss only changed the third decimal place of the total U value, and the tolerance in thermal resistance for all the elements is greater than that.  The same may well not be the case for a slab with very poor perimeter insulation though, and there are a lot of those around.  In our case, the entire perimeter of the slab has 200mm of EPS outside it, and 300mm of EPS underneath it, so the perimeter loss, even using the worst case thermal conductivity figure for concrete, and assuming that the edge of the slab would be the same temperature as the centre, was tiny. 

 

If you want chapter and verse on it then get a copy of BR 497 (it's expensive, though!) as that details all the BRE conventions for modelling heat flow for just about any possible building related component.

 

 

[PeterW]

6 minutes ago, JSHarris said:

The same may well not be the case for a slab with very poor perimeter insulation though, and there are a lot of those around.  In our case, the entire perimeter of the slab has 200mm of EPS outside it, and 300mm of EPS underneath it, so the perimeter loss, even using the worst case thermal conductivity figure for concrete, and assuming that the edge of the slab would be the same temperature as the centre, was tiny. 

 

I did this albeit with traditional slab. The PIR under the floor is edged by a 25mm expansion/edge insulation upstand however the inner core of the cavity is filled with 125mm EPS and this goes below the level of the slab to the top of the trenchfill. 

 

I did model this and it  has an impact on heat loss but not as much as you would think. 

[JohnW]

Thanks all for the explanations. I think I now understand that the P/A ratio has an impact on the overall U-Value and it sounds like potentially a negative impact in many cases.

How does the P/A ratio take into consideration the fact that all our external walls have thermal blocks for the first 2 courses and 25mm insulation to mitigate horizontal thermal transfer?

 

In my example, I know the P/A Ratio and the U-Value of the floor build-up but I would really like to know how the P/A Ratio reduced our overall floor U-Value from 0.17 to 0.14.

It seems to me that there are many variables at play e.g. soil type, which could allow a more cynical person than me to achieve whatever result they want by simply tweaking those variables.

[Jeremy Harris]

8 minutes ago, JohnW said:

Thanks all for the explanations. I think I now understand that the P/A ratio has an impact on the overall U-Value and it sounds like potentially a negative impact in many cases.

How does the P/A ratio take into consideration the fact that all our external walls have thermal blocks for the first 2 courses and 25mm insulation to mitigate horizontal thermal transfer?

 

In my example, I know the P/A Ratio and the U-Value of the floor build-up but I would really like to know how the P/A Ratio reduced our overall floor U-Value from 0.17 to 0.14.

It seems to me that there are many variables at play e.g. soil type, which could allow a more cynical person than me to achieve whatever result they want by simply tweaking those variables.

 

25mm of peripheral insulation will help a tiny bit.  Thermal blocks still have a higher thermal conductivity than insulation, so don't wholly mitigate the heat loss.  Adding perimeter insulation that extends below the slab level helps a fair bit, by increasing the heat loss path length as well as decreasing linear heat loss radially outwards from the floor.

 

You're right in that there are a lot of variables that impact the real-world performance, but the BRE put together BR 497 to try and produce a standardised way of modelling heat loss paths that is "good enough".

[ADLIan]

As stated the heat loss from a ground floor depends upon the size, shape, edge conditions and soil type. Best analogy I've heard is consider a ground floor like a bowl of hot soup - it will be cooler at the edges (=greater heat loss) and warmer at the center (less heat loss).

 

This table from BR 443 may help - larger, square floors having a lower inherent U-value.

 

 

With current levels of insulation in the general floor area (>100mm PUR or equivalent) vertical edge insulation has little effect on the overall U-value but is important to minimise the linear thermal bridge at the floor/wall junction (which is where BR 497 comes into play as guidance and standard conventions when using BS EN 10211).

 

 

[Jeremy Harris]

6 minutes ago, ADLIan said:

As stated the heat loss from a ground floor depends upon the size, shape, edge conditions and soil type. Best analogy I've heard is consider a ground floor like a bowl of hot soup - it will be cooler at the edges (=greater heat loss) and warmer at the center (less heat loss).

 

 

However, with sensible design the edge loss can be reduced such that it's so small as to not be worth considering in the overall U value calculation.  With sufficiently high levels of insulation underneath and around the perimeter there's no real merit in bothering to try and semi-3D model the heat loss, as just considering the simple vertical heat loss path gets to within around the third decimal point of the calculated U value.  This is especially true if the geometric bridge at the floor - wall junction is mitigated by ensuring that the perimeter insulation is both wholly contiguous with the wall insulation and of a similar overall thermal resistance.

 

Taking our floor as an example, the slab is cast onto 300mm of EPS, with a 200mm thick EPS upstand around all edges that has most of the thickness of the 300mm thick cellulose wall insulation resting directly on top of it (albeit with a thin layer of DPM between the two).  The outer edge of the wall insulation is 10mm in from the outer edge of the EPS upstand, but extends inwards over the edge of the slab by ~110mm.  With the edge of the slab effectively "wrapped" in thick insulation on the top, edge and bottom, the perimeter heat loss is pretty small, and not worth the effort of calculating in terms of the impact it has on the overall floor U value.

[ADLIan]

There are 2 issues here;

 

1. Calculation of the floor U-value, as above, with little benefit from the vertical edge insulation in the calculation

2. The linear thermal bridge at the floor/wall junction (along with may others) which is an input into SAP 2012 (& 2009). Default and/or accredited psi-values are available for 'standard' details however these would penalise a 'non-standard' but well designed floor edge detail such as the JSH detail.

 

With high levels of insulation poorly designed junctions can have a major impact on the SAP rating & Building Reg compliance.

[Temp]

2 hours ago, JohnW said:

How does the P/A ratio take into consideration the fact that all our external walls have thermal blocks for the first 2 courses and 25mm insulation to mitigate horizontal thermal transfer?

 

I haven't read the BRE document mentioned above but there seems to be layer on layer of BS and ISO standards that cover this whole area.

 

Googling eventually led me to ISO 13370 which in turn says that ISO 10211 has a "numerical method" for calculating the heat loss through the perimeter, and that ISO 14683 has a method using "tables" for doing the same (I think). I'm afraid I lost the will to live when I got that far.

[Jeremy Harris]

The BRE recommendations just simplify and standardise the 3D modelling process, to produce a set of recommendations as to how thermal models can be effectively standardised in order to all give essentially the same results for a given set of conditions, in this case compliance with Part L, primarily.

 

When reading through BR497 it's clear that the aim is to produce a set of rules that can be followed to ensure compliance with building regs, rather than a set of rules that produces a very accurate thermal model.  IMHO that's deeply flawed as an approach, as it results in details that may meet the needs of calculating thermal bridging as far as the regs go, but which take no account of the knock on effect of things like interstitial condensation.  There's a (rather bold) assumption that water vapour cannot penetrate a building from outside at a faster rate than it can escape, which is patently false for some construction methods and dynamic conditions.

 

I've yet to see an interstitial condensation model that can, for example, model the dynamic condition change such as a cold, damp, night, followed by a warm, hot morning, not an untypical wet of weather conditions in the UK.  When I did a simple 2D model of this sort of change in conditions over a period of around 4 hours I found that the standard SIPs sole plate detail resulted in condensation forming under the outer 30% or so of its bearing area, and there not being sufficient energy available through the rest of the day to drive that condensation  back into water vapour so that it could permeate back out.  A few such cycles would be likely to result in sufficient moisture build up as to allow fungal decay.

 

The solution is relatively simple.  Adding a layer of additional insulation on the outside of the SIPs panel, that overlaps for a distance below the sole plate, ensures that the outer edge of the sole plate doesn't cool down to dew point for any likely range of external temperature and humidity conditions. 

 

One of my main criticisms of the BRE is that they seem not to take a joined-up approach when looking at either modern methods of construction or dealing with changes to specific building regulations.  In the case of thermal modelling it's considered in glorious isolation, as they don't seem to have a systems philosophy when it comes to construction.  Arguably, this "consider each element in isolation" approach, which seems endemic in the construction industry, needs addressing.  Grenfell Tower is a fair example of how the absence of a joined up approach to materials certification and the application of those materials in construction practice leads to unnecessary deaths.