Posted By: Doubting_Thomas

Posted By: lineweightThe conductivity of steel is such that we can pretty much ignore the length of any heat path through the steel, or the cross-sectional area perpendicular to the heat paths.

I don't quite get what you are driving at in this statement but what you are describing generally is the difference between what are called linear (psi value, W/m.K) and point (chi value, W/K) thermal bridges.

There is a distinction in units, but also in the way that they are remedied. A linear thermal bridge can usually be resolved through adding other layers in the wall build-up, whereas a point thermal bridge will usually, by necessity penetrate through all these layers (unless thermally broken with a structural pad as described in the posts above) and is thus often harder to design out.

Others can probably comment on their thermal equivalence or otherwise (I'm not convinced), but I'm struggling to see why you'd want to describe them in the same way when they aren't likely to be used for the same thing in reality.

I'm picking up on comments from WillinAberdeen further up the thread, where he (I think) is making the point that a material such as steel has such a high conductivity that we can effectively ignore things we'd otherwise take into account. For example, just looking at simple U-value, for an insulative material doubling its thickness would have a significant effect, but doubling the thickness of a plate of steel would not have a significant effect on the rate of heat loss.

If I understood properly, I think WillinAberdeen's point was that it's the surface area of steel in contact with air either side of the thermal bridge that becomes the most significant factor. Much more significant than the volume of steel that all heat paths have to travel through in order to get from the "warm" surface area to the "cold" surface area.

Hence my thought experiment above, which is my way of testing that I have understood this correctly (and I might not have). Yes we might commonly measure one of those as a linear bridge and one as a point bridge, but I'm asking whether it's correct that we'd come up with a similar result for each scenario (measured in total heat loss for the entire assembly).

I think that in the case of the wall with the beam running parallel with the brickwork, we could calculate it by working out a U-value for the brickwork portion, and a U-value for the steel-beam portion, and then multiplying by the relevant areas.

In the case of the wall with the beam perpendicular, we can work out a U-value for the brickwork, but we can't do the same for the steel beam, we need to work out a chi value, but this is much more complicated to arrive at (and I think basically needs some kind of finite element analysis modelling).

And this comes back to the original question/problem of my thread, because if I'm trying to make a broad judgement on how to deal with steel columns penetrating an insulation layer, it's not a simple calculation, it needs some quite complex modelling to quantify the heat loss and how much difference a certain insulation strategy will have. Therefore me (not having the resources to do that kind of modelling) poking around for a way to decide what's "sensible" to do.

Posted By: GreenPaddyLineweight, does the simple calculation I suggested above not give enough information ..."to decide what's the sensible thing to do"? Pursuing a super accurate answer to 3 decimal places with a fancy model doesn't feel like a sensible use of time.

The quantity of steel outwith the thermal envelope only matters in that the more steel, the closer the steel will stay to the adjacent outside air temp. ie. lots of steel, and it will not be influenced by the heat coming from inside. But the heat flow would, in simple terms, be fixed with an assumed (guessed) temp differential, creating the energy driving force.

Yep sorry I didn't respond to that earlier, I guess I was hoping someone else with a better grasp of building physics would respond with comment on whether it made sense!

You basically said:

*******
Simple formula of ... heat flow = (k x Area x delta T) / length of flow path

heat flow (W) = 50 (W/mK) x 0.005(m2) x 10(oC) / 0.6(m)

assuming 200x150 "H" steel post of 10mm thick plate steel
assuming ground 10oC and inside 20oC
assuming 600mm from post base to above insulation

heat flow would be = 4W
******

As I understand it, the area you are using here is the cross-sectional area of the steel column, is that right? And then the "length of flow path" is the length of the column insulated from either the indoor or outdoor air.

So the amount of heat I am losing is proportional to the cross-sectional area and the length of heat path.

Double the cross sectional area, and the heat loss is doubled. Half the length of flow path, and the heat loss is also doubled.

This ignores how much column is projecting into the room or out into the cold air.

That is how I previously would have visualised it, but my understanding of WillinAberdeen's comments, is that this is not right.

I think that's because the amount of heat conducted at the surface boundaries between the steel and air, on each side, is really the limiting factor. So the 4W you came up with; this is the heat flow ignoring those surface boundaries, so would be the heat flow with a kind of infinite projecting length of steel either side, or something like that, but in fact the actual heat flow will be less, and heavily determined by the amount of steel surface area in contact with the air.

How those numbers work out though, I don't know.
0.5m of projecting steel means it's actually 3.8W and 1m of projecting steel means 3.9W?
or
0.5m of projecting steel means it's actually 2W and 1m of projecting steel means 3W?
or
0.5m of projecting steel means it's actually 0.2W and 1m of projecting steel means 0.4W?

I wrote a long reply last week and the forum software crashed before it posted :-( here is what I remember of it:

Love the thought experiment. The first example reminds me of sticking a skewer through a potato to make it bake quicker (worth trying IRL!)

The second example is like having 1920s steel window frames, bridging from inside to outside without thermal breaks. They definitely sweat, and many people seem prepared to spend £ now to specify new window frames to be thermally broken instead of plain steel, even if the total Watts lost is not very many. But that's a cost/benefit for the client/project to decide depending how high-spec the project is (as your OP question and GP's reply).

Seems to me that the most significant heat transfer resistance (insulation if you like) is the surface film resistance Rsi/Rso of the steel where in contact with the air. The two examples have different amounts of area exposed to air on the top/bottom surface.

In the 'skewer' example the flanges of the I beam act like the fins you get on the back of steel radiators. Note how thin those are IE they aren't bothered about the thermal resistance or cross section of the steel, so long as the air contact is good.

Cladding the exposed steel column in concrete would help cover up the 'fins' (reduce air contact area) and you could wrap/stick some insulation round the concrete at low cost?

Stainless bolts already mentioned. edit: much less thermally conductive than mild steel.

See e.g. https://www.schoeck.com/download/6616/SCI_Thermal_bridging_in_steel_construction_Schoeck_Isokorb_type_KS_and_KST__6616__.pdf/en-gb or https://asset.source.thenbs.com/api/pdf/5a4d5455-6c96-4cb4-ae59-88093293a087