Just wanting to get clear on the proper order when putting down vapor barrier and foam under a radiant slab. Would you recommend that the vapor barrier be placed on top of the foam (XPS) or under it?

I put my vapor barrier directly on top of the capillary break gravel, then foam above it and so on. Here are my reasons:
1) You want to keep your foam dry as it will absorb moisture.
2) If you put the vapor barrier on top of the foam, it will likely get a lot more holes in it as construction continues than if it was placed under the foam.

In some areas limestone is used which is quite sharp instead of granite so the holes will come regardless. I agree tho, VB under the foam.

Joe Lstiburek, principle at Building Science in Boston recommends putting the poly directly under the cement and above the foam. Given the potential for water saturating the layers under the slab, one wants the foam to be able to dry out by draining into the stone. Putting the VB under the foam prevents this and can result in the foam sitting in a puddle. Sounds like a valid argument to me, so that's the way we do it. Thinking that a layer of poly will, despite joints and potential holes, stop rising water is wishful thinking.

Posted By Bob I on 09 Apr 2013 09:53 AM
Joe Lstiburek, principle at Building Science in Boston recommends putting the poly directly under the cement and above the foam. Given the potential for water saturating the layers under the slab, one wants the foam to be able to dry out by draining into the stone. Putting the VB under the foam prevents this and can result in the foam sitting in a puddle. Sounds like a valid argument to me, so that's the way we do it. Thinking that a layer of poly will, despite joints and potential holes, stop rising water is wishful thinking.

The primary reason for putting the poly on top of the foam rather than below has more to do with short-term issues than the longer term issues of soaking the foam. 

With the poly under the foam, when the concrete is poured there will be some amount of water trapped between the foam & poly, which may take many months (or sometimes years) to dry. If there's too much trapped water in there at the time the finish floor goes it can damage the finish as it dries through the vapor-retardent foam.  In basement slabs that will never be finished (or won't be finished for years) it simply doesn't matter.  With the poly directly under the concrete any excess water is still in the concrete.  Water vapor can move through 4" of concrete more than 100x as fast as it can through 2" of XPS foam, so what takes 2 years to dry through the foam is gone in a week when drying through concrete.

Clearly the solution is to put in two 3 mil sheets - one on each side of the foam :-).

which proves that compromise does not always make sense or work

Posted By Bob I on 09 Apr 2013 11:06 AM
which proves that compromise does not always make sense or work

Whaddayah, some sorta congressional partisan pig or sumthin'?

Most do not address radiant floors when talking about radiant slabs. It makes no sense that a concrete slab would "dry" to the ground. One would have to assume that the ground is drier than the air above the slab, which is permeable.

This would be the equivalent of putting a vapor barrier between the insulation and the sheet rock of a finished basement.

We have known for some time that radiant floor dry out slabs. Many of ours run year-round to lower rH in otherwise cool, damp, basements.

Let's stick to the tried and true, compacted (therefore not "quite sharp") class 5 or equivalent, 10mil vapor "retarder" and 2" of EPS or XPS with PEX and wire as needed.

No sand anywhere, by the way.

Dana1: no I am not a congressional pig, but partisan? Of course!

I think there are enough pinholes that occur in vapor barriers to allow drainage of liquid water, that might be trapped there, yet still allow the membrane to do its job as a vapor barrier.
I don't agree with Lstiburek on this one.

Yabbut, Joe L has pictures of finish flooring damage caused by the trapped puddles under the foam. (They're somewhere on the BSC site...)

It doesn't take a very large pin-hole-free area to make a significant sized reservoir, but it does take a very VERY long time for even a 1/4" deep puddle to dry through 10-mil poly, and only about 1/10 that time (still very long) to dry through 2" of XPS, yet 1/10,000 the time to dry through 4" of concrete. Putting the poly atop the foam avoids the puddle altogether.

I am not sure where the "puddle" came from in the first place?

Putting the poly atop the foam avoids the puddle altogether.

Until there is some kind of leak or flooding somewhere in the basement. But that never happens, right?

not necessarily in the basement, but under the basement. sometimes the footing drains aren't that far below the floor (if there are any and if they're working). its not that far from possible that water could rise to 6" under the floor, where the foam is.

And Peter Pan can fly.

You can't build for any eventuality. Not with my money anyway. If the ground is not suitable for slab-on-ground construction then radiant heat is the least of your worries. A perimeter drain would take the pressure off, regardless of where you put the vapor barrier.

Or just be careful what flooring you use over a potentially moist slab.

finish flooring damage caused by the trapped puddles under the foam

Lstiburek has a document (BSI-003, Concrete Floor Problems) in which he expresses his dislike for the practice of putting sand under the slab and he does mention the vapor barrier, but the pictures seem to be of damage caused by the sand under wet slabs. Elsewhere he does mention his preference for placing the vapor barrier on top of the foam sheeting, and goes on to reference the BSI-003 document, but I didn't see where he fully explains his reasoning. If you are afraid that the vapor barrier will trap water you could always put a few extra pinholes in it on purpose.

I wonder if there is a difference between radiant and unheated slabs in this respect.

I have been on several projects where the concrete crew used their rakes to punch holes in the vapor barrier to let out trapped air.  Trapped air under the vapor barrier can let the vapor barrier rise up and make the concrete slab thinner in spots especially when the vapor barrier is simply thin vinyl.  I reviewed one project that the slab was so thin in spots that another slab should have been placed over it.

I may be wrong but because of the previous experiences, I prefer to place thin vapor barriers under the insulation.  I also prefer clean, tamped gravel instead of sand under a slab to raise it above existing grade.

Always makes sense to poke holes in vapor barriers; those things are a PIA.
Seems to be a rash of that going on here. Why not simply eliminate them & put up with the resulting moisture; wouldn't it save all that poking time?
And while we're at it, lets eliminate insulation - that would solve the mold problems.