Tricky to explain so bear with me.
We're extending the basement of an existing listed building, the image below shows a simplified ground floor (left) and basement plan (right). Blue shows existing footprint and green shows new areas.
Due to listed status our options are limited so it's IWI to all walls. New build basement walls will be retaining so generally we're maintaining the same approach here with solid masonry walls and IWI, any EWI would likely be crushed. Yellow walls are lined out to achieve the same U-value (0.28 W/m²K) as above ground.
What I'm wondering is how much we need to insulate the new basement walls shown in red. They are not 'perimeter' walls in the sense that they will not be exposed to outside air, as they are completely beneath the ground floor footprint above. Following FosterTom's theory that the soil directly beneath a building will reach an equilibrium point if continuously heated, I'm wondering what our options are for these walls.
It would be ideal if we could reduce the IWI depth here as space is very restricted. I concede that the perimeter 'enclosure' for the soil here is just the existing building's foundations, so won't be insulated, but is there still a case for reduction?
Finding it difficult to find ways to model this scenario, although I'm gradually getting to grips with THERM if anyone can give me some pointers?
Comments
The red walls are solid reinforced in-situ concrete. Basement is around 2.5m high internally.
But how do you stop BC making you install a vapor barrier on the inside of the insulation?
It's a cavity drained system so any hydrostatic ingress should be dealt with behind the lining. Tricky part is what type of insulation to use in front of this as we have woodfibre above ground but I'm not sure it's suitable below DPC.
Tony, we can't go above 0.28 on other walls for this reason - WUFI says OK to achieve this but anymore would risk the dew point.
I guess I need to get a BCO to confirm what U-value they would accept first, I was just hoping to build an argument before getting them to name a figure.
BTW, and off-topic, I used some EPS that my client had saved from a skip. It had lots of pibk bits in, and they were *sharp*! Anyone come across this?