Hi all.

I have a question: does anyone have a good example of calculating a thermal bridge up against another building? Meaning for a partywall or connection between one building and a neighbor?

I am confused about exactly how to do that calculation - with my current method I seem to get a very different Psi-Value depending on if I assume the neighbor's wall is just bare brick or an insualted wall? It seems to me that shouldn't factor into this calculation, should it? (or rather.... should get factored *out* somehow)

**Plan view up against a neighbor's wall. Exterior on right, Interior on left and top. If I assume the Neighbors wall is uninsulated **(*which might be true?*):

**Or if I model their wall insulated **(*which might also be true - who knows, its not our building...?*):

seems I might be doing this wrong if I'm getting different results like that? Is there a known protocol for details of this sort that we should be following?

Any thoughts or examples would be much appreciated!

thanks very much,

-Ed

Hello Ed,

I'm jumping in here with my 2 cents, not sure if this has been covered already (sorry for the repeat in case). Also, I'm not too familiar with Flixo - we use Dartwin Mold simulator and while they both comply with the same standards, the intermediated metrics all the way until the very end are structured differently from what I see here.

What you have here is a thermal bridge with more than two boundary temperatures. This is covered in ISO10211, paragraphs 8.3 and 10.2.

Long story short, the architecture of the calculation depends on what you assume to be the neighbor temperature - if you consider it to be at the same temperature of your building (i.e. adiabatic), or if you want to be conservative and assume the neighbor to be at a lower temperature (say 64˚F?).

Another thought that may influence the resulting PSI value (i.e. PSI = L2D - U * L * H, with H here = 1 m), is the length you associate the external wall U value with. Both models show L = 1.20320 m, but that is incorrect.

This depends on how you are calculating your gross building envelope volume in PHPP. If you 'parting' your building from the neighbor one along the mid line of the party wall, you need to extend the reference lenght L to that mid line. That will change your PSI value.

On a high level, it looks to me that the neighbor wall being uninsulated in the first example leads to the masonry wall being 'heated' by the neighbor, for the benefit (lower heat losses) of your building.

With the neighbors insulating their walls, their heat losses are cut to a quarter (from 41 W/m to 11 W/m). This causes your building to take up a larger share of the 'heating' of the thermal bridge, with the PSI value going up accordingly.

I mean, somebody may see the neighbors insulating their own side of the wall as being passive aggressive but hey, we're all saving the planet right?

Hi Ed,

I've been doing some fairly complex 2D thermal models with multiple boundary conditions lately, like these ones below, and having to come up with modeling protocol for certification and to teach to new CPHC's. Not exactly party walls, but similar concept and boundary condition issues:

I'm happy to jump into a meeting with you to help you figure out what's going on if my answer below isn't clear, but from your examples above here's what I see:

ULdT of 2D model - (sum of ULdT's for all flanking component clear fields) /dT

Maybe first double check you have the correct ULdT clear field for the insulated party wall, and check if your lengths are correct. (this would make a negligible difference though). Feel free to email me [email protected] and I'm sure between us we can solve it!

Cheers,

Prudence

Hi Ed,

I use Flixo and found this document( I downloaded from somewhere) very useful. The author describes a thermal bridge as "...both a correction factor and an indicator of heat flow but

primarily a correction factor" . So when the R value of the adjacent walls go down the correction factor (Psi) goes down... and when the R value of the adjacent walls goes up the Psi (correction factor) of the constructions joint goes up. I think he is saying the Thermal Bridge is not so relevant when you are loosing so much heat from the walls anyway.I have noticed this when reducing insulation in 2D TBs to only get a lower Psi figure. I can't screen shot his illustrations because of copyright but happy to send the whole pdf as I freely downloaded it a few years back) Look at p35 onwards. The authors are Jacobson & Petit (I assume from a presentation), called The Why of Psi for the Certifiers Cooperative...I will send the PDF to the Accelerator hello email.

It may not answer your query

regards Torquil

Yes - this is counter-intuitive for many people, but heat flow will just continue to take the path of least resistance, so thermal bridges are worse (higher) when flanking components are more insulated

Thanks all! you are all awesome. I really appreciate the feedback and suggestions. Those are all really helpful pointers.

I am pretty sure we're correct on the lengths and U-values here as the separation line between our building and the neighbor in our PHPP is drawn right at the outer surface of our new CMU wall there (not at the center of their exg wall) so I'm mostly certain thats all correct. Its definitely a funny condition for sure - the proposed construction sort of looks something like this:

if that makes sense? So our energy model includes an adiabatic wall up against the neighbor, and then we have our new insulted wall along the exterior. So roughly like:

So I would have thought that the Psi-Value here should take into account the extra heat loss due to that terrible CMU bridge at the corner, as well as the extra heat flow along what we'd originally tagged as the 'adiabatic' surface (which, in reality, has a little bit of heat flow due to the bridge) but then factor

any heat flow from the neighbor. But I suppose as you point out, thats impossible isn't it? We have to make some assumption about the neighbor's wall as their HF will affect ours....outOr could you ever just do it without the neighbor at all and call that entire edge simply Adiabatic?:

It seems as though that would be the most conservative solution since you'd be completely neglecting any beneficial accidental heating from the neighbor, right?

at any rate, thanks again for all the feedback! I'll keep working on it....

best,

-Ed

Hello Ed,

Thank you for the additional notes. With your explanation, I see your model dimensions being correct.

If you look at the temperatures along the existing wall + CMU, you can see how much further in the 'cold' penetrates if the neighbor insulate their wall. In the screenshots, I'm pointing out the red isotherm just for reference, assuming it refers to the same temperature in both screenshots.

This results in more area of the party wall of your building being affected by the 2D thermal bridge, aka higher PSI value.

As far as considering the neighbor building adiabatic, I think what you are describing makes sense only under the assumption that the neighbor building is going to be continuously heated, at a temperature equal to or higher than your building. What if the building is unoccupied/unconditioned for some time? Me being conservative, I'd consider at least 2K temperature difference on the party wall.

Regardless whether or not you consider the neighbor adiabatic, the heat losses from their building into the existing wall shown in the junction (brown) does affect the junction heat losses of your building.

Hi again Ed and Enrico and Torquil.

Fun- discussion!

If it were me, I'd also set up a worst case: (ie uninsulated party wall and unheated neighboring building), so BC with interior horizontal heat flow for air film and I agree with at least 2K, maybe even 4K temp diff. I'd try to find out about the neighboring building envelope and do a quick calc to get estimated interior surface temp (without the party wall).

In THERM I'd experiment setting the U-factor tags (or 2D detail flux "gate") on both interior sides of the party wall AND as it looks like you have done, on the exterior side, and see if that yields different results (to satisfy myself results are consistent), making sure to adjust lengths so as not to skew my results.

Let us know where you end up!

-Prudence

Thanks for the input! I like that idea of modifying the neighbor's interior temp - it seem like an especially conservative way to approach it from a moisture and condensation risk analysis perspective.

So I went and asked a colleague from Passivate about this as well and his suggestion was that I should be doing it as a '2-Construction' Psi-Value with one of those being the neighbor. So like this:

whats interesting using that approach is that no matter what I change (Prudence as you suggested: testing with changes to lengths of elements, insulating the neighbor, changing boundary condition temps) it always ends up more or less in the same Psi-Value (ish). Thats seems like a really good sign to me that I get the same result no matter how the initial simulation is set up. For instance:

Add 100mm of Mineral Wool Insulation to the Neighbor's wall:Make everything twice as Long:Change the Temp of the Neighbor's Interior to 17°C:And very interestingly (to me at least), even with full insulation thickness right to the Neighbor, we still have a pretty large Psi-Value:

Which totally makes sense to me since we've got heat flow 'around' our wall through the neighbor's.

What do you think? I think this seems much more believable. I guess the other question would be: Would you divide this Psi-Value in half? Half the energy is coming from the neighbors and half from us after all right? So like a window mullion where you'd take half the Uf for each frame: split the heat loss here in half for purposes of the modeling?

regardless, thanks again for the thoughts and feedback!

-E