Near outside wall 750mm down inside house = 16.1 C this is 600mm and 600mm from an external corner but under an unheated cupboard

350 down = 16.2 200 down =16.7 100 down = 17.0

It is a bit confusing it being under inside a cupboard bit the cbd is at 18.5

What's the wall thickness/material, Tony?

Have you checked the floor/subfloor temps lately Tony ?
tom

come on guys n gals, get tony back on this, and please help mike7. Info like this could be our saviour

tom

I am also v interested.

Many thanks for your whispers, Peter and Paul - I probably should wait until I'm a bit more sure what I have is somewhere near right. I keep finding errors but it's getting better - when the results confirm my expectations, I'll know it's right.

Howdytom, don't get too hopeful ...

There is a stong smell of hot wood coming from my abacus.

Brilliant !!!!

In other words the current ideas about insulating solid ground bearing floors are misguided and edge and perimeter insulation should be installed instead.

thanks for all that hard graft Mike.

tony

Wow is that what it means? Good timing - time to study it properly.

Blimey, is that what a 50kPa test-wind does to your walls?

If insulating externally, run the insulation down to top of strip found. That's what I'm detailing currently on a bungalow refurb and I'd appreciate feedback:
As it's refurb, a narrow trench has to be dug, below-ground insulation fitted and backfilled immediately.
Trenching hopefully cheaper as no H&S shoring - no one goes down the trench.
Would hardboard be suitable as one-off protection against backfilling - or something even cheaper? Some say bubblewrap!
Anyone know if either ordinary or Platinum-type EPS is not only durable in assumed-permanently-wet conditions, but will also stay non-waterlogged and so retain its insulativeness? Or must closed-cell XPS be used to resist waterlogging?
On a bungalow refurb of this era, top of strip found is likely to be only 750-900 below FL, assuming FL 150 above GL and found depth 750-900 below GL, as was then considered adequate (1000 now). So that's not a great downstand. What if a little bit more 'wing' insulation were added by coming out horizontally across the strip found top, as wide as the trench allows?

I'm also considering convincing the Building Inspector to follow same strategy for the newbuild extension bit - no or very small insulation under slab, 150thk like this as perimeter downstand.

Someone kindly ran off these figures, from some software he had:
14m x 14m floor plan
Uninsulated slab effective U value 0.52 (i.e. U-value averaged over the whole area, but obviously much greater near the perimeter, very little near the centre)
with 700 deep downstand below FL, 100thk EPS 0.42, 150thk 0.36 (and much more consistent across the whole area, because the edge loss is reduced in importance)
with 1000 deep downstand, 100thk ?, 150thk 0.33
with 1400 deep downstand, 100thk 0.32, 150thk 0.30, 200thk 0.29
with 1400 deep downstand, 100thk, plus 25 under slab 0.26.
The last figure is interesting - the great U-value improvement offered by just 25 of insulation. By biassing or 'tuning' the temp gradient, it also has the effect of disproportionately raising the temp of the floor temp, for comfort purposes. However it does begin to significantly cut off the underfloor mass from immediate contact with the interior, I'd have thought?

Opportunity for a chortle doesn't mean poor!

You mean ordinary EPS not closed-cell XPS, below ground? and does twice/three times as much mean 300/450thk? The flaunch says that underground downstand insulation needs to be thicker than ditto above ground?

A little progress with the simulation: the 'poor man's 3D' version is now running. Instead of showing a cross-section of a very long building (2D), it calculates for the solid of revolution generated by rotating the section about a vertical axis - so it gives answers for buildings with a circular floor plan. Not quite perfection (!) but much more realistic than the 2D version, and a lot less bovver than proper 3D. Should be quite OK for heatstore calculations, where the heat patterns will quickly approach spherical/cylindrical after a short distance.

I've done nothing yet with your floor/wall design suggestions - thanks Tony and Tom - but have been playing quietly in my corner checking my results against Tony's measurements on his actual house. Not a perfect match, of course, but close enough for me to have a little confidence.

I equated Tony's building to a round one of 4m radius with roughly the construction he told me. The grid is 0.1m, so to simulate 50mm of cavity insulation I used 100mm of material of twice the likely actual conductivity. The yearly temperature cycle of 10degC +/- 6C is about right for Oxford - the nearest I could find to Tony's location. Possibly a touch low.

Posted here is the sheet for November with a ground conductivity of 0.6W/mK as per Tony's assessment. Also month by month graphs of heat loss per m2 for the wall, perimeter floor, floor minus perimeter region, and whole floor - for the 0.6 conductivity, and also for k=2.0 such as you might get with wet clay, for example.

Strewth - I'm really having trouble with posting this stuff - can't even remember what I did last time... Sorry it looks so crap. Not much point if you can't read it.

This is very intersting stuff. I've got a house with similar floor construction to Tony's (but newer - 1960). 20m wood block, bitumen glue layer, 40mm screed, weak concrete, depth not yet determined. Floor is 7m x 8m rectangle. I could be persuaded to drill some holes and see what's what. Heating here is intermittent in last 2 years, mostly woodburner and it gets down to 13C if you leave it for a day or so, so I suspect my underhouse soil will not be much warmer than the annual average. Ground is chalky, but I'm not sure if water table is much more than a couple of metres down, which of course prevent this scheme working well. Still if we had an example where it did work and one where it didn't then that would be instructive for when this is/isn't a good idea.

Ultimately I need to decide whether to insulate floor or not. I've been assuming to date that it's a good idea, but this thread allows for some doubt.

Mike can you put up your new 3D sheet and give us a link to it?

on three sides of my house there are gardens -- 4 th side is another heated house kept warm pretty much 24/7 it is 5 m away and between the two houses there are unheated garages -- these would help to create a heat island effect under our two homes.

i m using a thermocouple and monitoring it with my computer

In my new house I will monitor up to 20 and log data from them all!

mike7, I just had a 10-minute go with the 2D sheet. I don't have Excel anything - I'm a GNU/Linux person. Loading it in Openoffice, it is largely full of err:523 'does not converge', which appears to be due to the recursive definition in M3 =(M3+1)*O3
(03 is 1). I'm not sure what spreadsheets do with this sort of recursive function? Maybe I need to tell it to recurse harder?

gnumeric seems to do a better job and doesn't complain about this, so it seems to be working - numbers look plausible. Macros are fatal to this kind of cross-platform use of sheets but you appear to have avoided those, so that's good.

Shouldn't the bottom row, which is the deep soil temp, come from $I$23 rather than just being '10'?