ned Productions - Friday 9 September 2026

10:34.

Word count: 3325. Estimated reading time: 16 minutes.

Summary:: A house build is described. Subsoil was removed by trucks. EPS for foundations was delivered. Underfloor heating materials were purchased for DIY installation. A covered roof area was designed using an AI tool. Satellite imagery and drone photos of the site were compared. Future installation of insulated foundations is mentioned.

It’s off to the races with my house build now that the soil pile has been removed:

Soil removal is obviously the right business to be in: I had quotes ranging between 5k and 14k to remove 350 to 400 m³ of subsoil. The cheapest quote at 5k, Green Valley, are unsurprisingly very hard to get to come out to actually take away your soil and I ended up ringing them weekly for fourteen weeks. But out of the blue the owner rings me at 6pm and says they’ll start taking away the soil at 8am the following morning, so I cancel everything and here is the log of trucks from my security cameras:

08:02 25t
08:33 25t
08:55 25t
09:10 25t
09:49 25t
10:43 25t
10:49 25t
11:12 18t
11:32 18t
11:36 18t
12:19 18t
12:23 18t
12:31 18t
12:51 18t
13:35 18t
14:13 18t
14:20 18t
14:39 18t
14:45 25t
14:49 25t
15:35 25t
15:41 25t
15:51 25t
15:57 25t

That makes for an approx total of 523 tonnes of subsoil removed, which if it weighed 1500 kg/m³, would be 350 m³. So about spot on. Unfortunately, Green Valley charged me rather more than quoted at ~€7.5k inc VAT, but it was still cheaper than the next highest quote at €9k so I guess it worked out okay. Just kinda sick of how expensive everything which uses diesel has become 🙁.

And what an unblocker the soil removal has become, because we have gained back the space to store bulky stuff. Less than two weeks later, the EPS for my foundations arrived from KORE:

That cost me about €18.5k inc VAT to cover about 240 m² of ground floor slab, which is €76.88 inc VAT per m² or €256 inc VAT per m³ of extruded polystyrene, with an expected u-value of 0.1025 W/m²K as according to KORE (we have yet to see what PHPP thinks!). Two years ago it would have cost €150-200 per m³, but as the styrene feedback which is derived from naphtha is now 25% more expensive, so is the EPS. Also, I paid extra to have all the EPS100 in silver (i.e. graphite enhanced) rather than white, which dropped the u-value by about 10%. It cost an extra €400 ex VAT, which is a bargain given the improvement.

As now the foundations are going to be installed soon, I ordered the materials for the underfloor heating, which I’ll be installing myself. That mainly consists of 500 metres of double walled 16mm diameter UFH pipe, the manifold and a luxury extra – castellated panels:

The castellated panels alone cost more than the rest of the UFH materials order, and I chose the cheapest ones I could find. What you do is cut them to shape and glue them to the EPS. You then thread the UFH pipe between the raised knuckles, which holds the pipe in place. You add the mesh on top, and clip and wire the UFH pipe onto the mesh where that makes sense so it won’t move when pressurised. You then pressurise the whole system to a couple of bar, and pour your concrete slab.

The professional installers don’t bother with castellated panels as they’re too expensive (I paid €11.34 ex VAT per m²), but for somebody like me who has never installed a UFH before they save large amounts of time and stress because they avoid problems like the pipe holding pins popping out when the pipe comes under pressure and you’re spraying concrete at it, which then means you need to react quickly and wade through the liquid concrete through which you can’t see anything to save the slab and the installation. The panels just make all that stress go away, and they also let you very easily ensure an accurate 100/150/200 mm spacing depending on your UFH water flow temperature. So yes a very straightforward swap of money for your time and stress levels, but one I reckoned worth it.

Wet underfloor heating is seen as one of the most expensive space heating systems, and it is if you hire in specialist installers. A bit like with the solar panels though it’s all labour time: the materials cost me €537 inc VAT and the castellated panels €720 inc VAT with 500 metres of pipe. For my system with a flow temperature of 35 C and radiation of 45 watts per m², you would use 150 mm pipe spacing which means 6.67 metres of pipe per m², so 500 metres will cover about 75 m², of which about 40 m² will go into the main house slab and the remainder will go into the outhouse slab. If 1,800 watts sounds ludicrously low for an entire house, remember due to this being certified Passive our total space heating is 2,600 watts, and in fact a good chunk of the total UFH piping (600 watts!) will be going into the floor of the conservatory because why not have toasty warm feet whilst being ‘outdoors’? So the indoors is only about 1,200 watts, about half the total space heating requirement – and don’t forget that the heated air ventilation alone can cover all our space heating needs. So the UFH is principally to reduce the ventilation rate and therefore heat losses and fan noise, not that we need it in the absolute sense.

Given that the going rate for professional installation is about €100 inc VAT per m², going DIY should save me about €6k. I’ll get this done for €1.2k, or €16 inc VAT per m² including the castellated panels – perhaps now you can see why the castellated panel luxury is very much worth it given the overall savings.

Covered wall area

I – or rather, my AI – has been pressing ahead on the design work to get the mobile home installed. There is a non-obvious problem with a mobile home: there is nowhere to put the washing machine or dryer, as those are usually provided by the camping site. We own both our washing machine and dryer so that part is easy, but where and how to install them onsite where they are accessible and can work is rather harder.

If you want some covered space from the weather, you have a few options:

Put something cheap in, and accept it’ll last a few years after which you’ll either need to replace or dispose of it, losing your investment in the process.
Put something temporary in which can be later resold, so you’re effectively renting it for a period.
Put something permanent in, that will last you until you’re dead, but now you have the problem of designing something which is good for multiple use cases over decades.

I’m never keen on Option 1, it’s wasteful, though the existing garden shed is a good example of it: cost me a grand and two weeks of my life I can’t get back. It’s been there for three years now and it’s held up well so far, though the screw washers on the roof have begun to split which means rain will start to ingress from now on, which means the roof sheets will begin to corrode. At 0.25mm thick they won’t take long to fall apart, once the protective paint layer is gone such thin metal corrodes quickly. And that’s okay: I expected ten years, it’s on track for that and you get what you pay for.

Still, it’s wasteful, and Option 2 of dropping a 20 ft shipping container there which you can later resell is probably the right option. My AI reckons that option would cost about €7k inc VAT including a base for it, and you should be able to sell it on DoneDeal after for maybe €2k. So €5k total cost amortised over however many years you keep it.

Which raises Option 3. There is only one place I can really place this storage, which is next to the existing electricity distribution box at the south west of the site between the shed and the lego concrete blocks – which is also where the mobile home is to be placed for the same reason. I’ll be installing services for the mobile home, so it would take little extra effort to extend them to the western wall to a concrete pad on which up to three 600mm wide appliances can be located with electricity, mains water and a foulwater available. However, that location in years to come is right next to where I want to place my fake holy well, so whatever design is chosen would need to complement that.

Which is hard this far out. I did originally plan to have a small covered roof area there for the soil oven used to sterilise soil across seasons – it’s why I fitted three phase electricity there. But I need a lot more storage than that right now, so one route would be to install a much larger than originally planned covered roof area which down the line could perhaps function as a workshop, or dining area, or even a potting area. The 2m x 1m concrete pad with drain could later have a large Belfast sink fitted, then it’s somewhere you can wash dirty outdoor things. Or maybe one could fit a mini bar there, right next to the fake holy well water feature, and that might be very nice. There are plenty of options.

We are width limited by the future fake holy well which begins where the lego concrete blocks are, also spans wider than 3m get expensive from a structural perspective, but we can extend the covered roof area the full length between the lego concrete blocks and the shed, which is about 8.9 metres. Assuming box profile roof sheets from nearby Irish Rollforming, they come in standard one metre wide profiles with 200 mm overlap, so total length must be <= (1000 + 800 * N) which is 8.2 metres. That felt excessively short, so I ended up accepting paying for an additional sheet 200 mm of which will need bending down, which makes a covered roof area 8.8 metres long.

But what combination of materials and structure should we use? As luck would have it, a couple of weeks ago there was a stealth LLM available for use for free of cost called ‘Space Bunny Alpha’ as they were testing some soon to be released new LLM, probably MiniMax M3.1 Flash. So I put it towards researching all the possible materials suppliers able to deliver to North Cork and their prices, and I asked it for a cost optimised design. This is what it produced:

Now, no doubt that could be better, the English prose in particular reads like a bad translation of a Chinese language original. But for an alpha quality LLM which was admittedly riddled with bugs and issues, AND it was free of cost, it’ll do. The most important bit is that its structural calculations are correct, and on that it did well:

It researched what climatic conditions are likely in my exact location.
It researched all the building codes for UK/Ireland, and then scrupulously enforced those.
I only fed it the survey from a few years ago and a drone photo and told it where to put the covered roof area. It did the rest by itself in terms of measuring and modelling things, including the scaling of aerial photo clip to survey so they match at the front of the PDF.
It realised early on that the existing wall isn’t strong enough to handle the turning moments a two metre plus wide roof can generate with a wind gust. It did have the idea of bridging off the existing solid concrete support piers instead, as those are much stronger.
It realised early on that the evergreen trees on the other side of the wall would produce continuous leaf fall, and therefore oversized gutters would be needed otherwise they would clog (which I can confirm from the gutter on the existing 1m wide covered roof area to the north). To do that, it calculated leaf fall rate, rain fall rate, and therefore estimated the gutter clearance rate. I was quite impressed!
It took longer to settle on the front eaves design, getting a bit confused about whether it wanted vertical or horizontal strength. In fairness, it was a bit complicated – the front eaves beam will take most of the load of the roof, with the wall rail providing secondary not primary support. There was a cost minimisation problem in there too – galvanised square hollow section (SHS) comes in 7.5m lengths, and it’s the critical limiting factor in the overall design as it’s the one material we expect to have zero of it left over afterwards. We are buying three lengths which is 22.5m meters, take off the two front poles 2.65m long and the 17.2m remaining turns into two 8.6m lengths i.e. this particular material constrains the whole design, it’s the critical limit.
It took many iterations and days of crunching to produce a full computer model and simulation which lets you tweak the design and it’ll automatically go cost optimise everything, including a cutting schedule for the materials to minimise wastage. But it did get there, and it chose galvanised steel for everything after evaluating aluminium, stainless steel, and timber alternatives.
A very comprehensive test suite including randomised mutation testing was produced.
And it did spit out a full architectural IFC file which you can investigate here:

For free of cost, I can’t complain, despite the many glitches, quirks and Chinglish prose of an alpha quality LLM. I did override its judgement on a few things:

I ordered it to cast the poles in place in concrete, which is more expensive than the casted in bolts with plates approach it preferred. But casting in the poles directly is much easier for me.
I ordered it to use thinner rafters than it wanted to save a considerable amount of money. These sag more than building code allows, but only by 15% and for the hundreds of euro saved I think it worth it (the fairly steep slope of ~10% means no water will pool from the sag, which is only 13 mm in any case).
It was completely incapable of doing fastenings between members for some reason. Those were easy to do by hand, but I found it interesting just how confused it got even when I told it exactly what to do – it couldn’t perceive what I meant. Which is probably fair for a coding agent not intended to make construction plans … but still, it was interesting to see where its perceptual limits hit a dead stop.

In any case, I expect this covered roof area to cost about €4k inc VAT of which €2.2k will go on:

4x 40x40x3 galvanised SHS 7.5m, for the rafters spanning 2.37m each.
3x 80x80x4 galvanised SHS 7.5m, for the front posts and the front eaves beam which is two 80x80x4 stacked vertically and connected by a 3mm thick plate both sides.
2x 100x100x5 galvanised SHS 7.5m, for the rear wall posts and the rear support beam (which we intentionally let over-sag to provide gutter fall)
1x 2500x1250x3 galvanised sheet, from which sections are cut to connect things e.g. we use 2500x160x3 strips to join the 80x80x4 SHS into a double height beam.

It’s a lot of steel, steel is heavy to transport and galvanisation has become especially expensive recently because it requires electricity, and electricity prices have surged in Europe recently. It is what it is in the end.

I’ll probably use Irish Rollforming box roof sheets on it like I did for my previous one metre wide covered roof section towards the north of the site. They’re absolutely standard size and profile same as any other supplier, they’re priced about the same, and they’re locally manufactured. Interestingly, the AI reckons 0.6 mm thick is the sweet spot – any thinner and the roof sags more as the metal is weaker, any thicker and it sags more from the added weight. So 0.6 mm thick it shall be, and I know from the existing 0.6mm thick section on my existing wall that at that thickness they are a fine weight – you’ll need two people to lift it on in any case. They should last pretty well – I have small polycarbonate roof lights in there too, you get up to twenty years out of those and I reckon the whole roof may need replacing before I die. The structure however, that should be good until I’m dead – we’ll simply unscrew the existing sheets and replace them when the time comes.

The total cost of around €4k makes this Option 3 covered roof area about €170 inc VAT per sqm. The next cheapest option of an Option 2 shipping container you sell later is about €370 inc VAT per sqm, and you get 66% less storage (though, because it has walls, all of its storage space is usable for all purposes). A self assembled timber carport is also about €370 inc VAT per sqm, and would be an Option 1 throwaway investment. It wasn’t that many years ago that that kind of cost per sqm would buy you a shell of a house, but those days are long gone now.

How good has satellite imagery become!

So that’s not actually from SkyFi, I had an unusually calm day last time I was flying the drone so I took it up to 200 meters high for its first time. I did consider taking it up to 250 meters, but my nerves wouldn’t take it – I absolutely cannot afford to lose this drone until the house build is done.

Mine, being an older model, will let you go up to 500 metres with just a warning after 120 metres – there is no hard coded maximum for these older models. Which then made me wonder? At 500 metres is that actually worse than SkyFi nowadays?

SkyFi will sell you a 14 cm/pixel image for the site’s location for US$35 nowadays – just stunning how far things have come since I last looked at satellite photos of the site in 2023, where the best available was 75 cm/pixel. That’s a 25x resolution improvement for only a bit more money (about the rate of inflation) in three years. At what height would my drone get an equivalent resolution?

Well, at 200 meters up I reckon the drone sees about 300 metres of view, which is a ~74 degrees field of view, not the 82 degrees in the spec sheet for the drone. Though it has a claimed 48 MP camera, in reality you get more like 24 MP at best due to the Quad Bayer sensor, so on that basis 6000 pixels over 300 meters would be exactly 5 cm/pixel. Therefore, to reach 15 cm/pixel, you would need to fly the drone at 600 metres altitude.

That’s nearly doable, I think my older model drone has a hard altitude from takeoff cap of 500 metres – and, to be honest, I’m not sure how comfortable I would be flying a 250 gramme drone at that height, it would be very likely to get blown all over the place by gusts in ways which could damage it.

Still, food for thought I suppose. Vexcel, one of the main commercial satellite image suppliers, captures at 7.5 cm/pixel already so it’s only a matter of time before that level of detail becomes cheaply available on SkyFi. That’s equivalent to a 300 metre altitude on my drone, nearly as detailed as the photo above. Nuts really.

What’s next?

Almost certainly pictures of the insulated foundations being installed. That should start next week – watch this space!

#house

Go to previous entry

Go back to the archive index

Go back to the latest entries

Niall’s virtual diary archives – Friday 9 September 2026

Covered wall area

How good has satellite imagery become!

What’s next?