I have a second floor addition that has 30" of space (and the old roof) between the first and second storeys.  This space is conditioned, but I'm concerned that, especially in the bedrooms with carpet, the heating will not be very efficient giving up it's heat to that inter-storey space rather than the room.

I've seen commercial product that combine, say XPS and OSB to create an insulated floor and I've seen XPS floor supports (without the OSB).  One of these would allow me to put insulation between the tubing and the inter-storey space, however, all these solutions are pretty expensive. 

Is there any reason not to roll-your-own?  Polyiso would also seem to be a better material, so I could glue polyiso to the existing subfloor, use a router to create gooves for tubes and then put 5/8 (7/16?) OSB glued/screwed on top.  Maybe 3/4 ply under wood floors (about half the space).

Anything to watch for? Any reason why this won't work?  Are the commercial providers doing anything special? Special foam, special testing??  Any perfomance data I should be watching for?  Polyiso seems to have a compressive strength of 15+psi, certainly more than 40 psf required by a floor.  I assume that EPS is about half that, but that should still be sufficient.  Anything the building inspector is going to object to?

don't know, but I can say don't omit plates. covering half of the tubing with foam won't help your output.

The plan was existing subfloor, insulation, tubes in top of insulation then OSB.

So, hopefully, the tubing would be in pretty close contact with the OSB.

Do you think that plates might still provide a significant advantage even though down is R12 and up is, say R3?

yes. contact area and quality is important. otherwise you just have a naked staple up with better insulation which is weak and not typically a low temp application, and significant heat striping concern.

Rob,

If I look at something like the Roth panels, they have a very thin layer of aluminum on the top, much thinner than a "standard" plate.  I've seen you mention these in other posts.  Do they work, is that thickness of aluminum sufficient in an underfloor application?

you're not looking at something "like" roth panels if that is your observation. Roth panel is thick gauge aluminum. Most overfloor panels are not: most use aluminum foil or very thin aluminum to seem like they are doing more than they are.

Roth, Warmboard, Raupanel, and Thermofin U systems are the only heavy gauge panels I know. If I wasn't going to use them, most of the time I would go with lightweight plate sandwiches before I used any other panel.

Thanks Rob.  I was trying to find specs on Roth, but didn't have much luck.  How thick is thick in this case, or how thick would the aluminum need to be to be effective in your experience? 

Relatively thin aluminum  (7mils) is readily available and easy to form.  I'm concerned that it's a little thin although double, triple (or more?) layers are easy enough.

hard to say. we use 11/1000ths in our lightweight plates, and some are up to 19/1000ths. Roth and the other panels I mentioned are significantly thicker than that.

tubing contact is one critical detail that is hard to get right with thinner plates.

So double-7 would work to give 14/1000ths. Maybe a sandwich. The aluminum has a little "give" and so the top might be used to push the bottom into better contact.

I don't think you could pay me to layer aluminum in a floor assembly. I would think the potential for noise would be significant?

Where do you think the heat "lost" to the dead space would go? As long as the ceiling of the room below has insulation in it, it's not going anywhere. Blow some insulation into that inter story space and not worry so much rather than going with more complicated & expensive. 6" of blown cellulose would give you about R20+. The heat loss of the extra 2' of exterior wall won't amount to a lot if it was insulated when the second story was added, if reasonably air-tight. The thermal mass of 20-30" of dead-air is miniscule compared to 3/4" of wood subfloor, so it won't much affect response time either.

Could easily be that my concerns are unjustified - at least I think that you've focused me on the nub of my issue.

There's no insulation in the ceiling below. The walls around the space between the first and second floor are well insulated and sealed, but the space itself is fairly open, but I'm not really worried about heat loss from that space. There's the remains of a tar-and-gravel roof between the first and second storey, so there is thermal mass, but I'm not really too worried about response times.

I think that my major (only) concern is the rooms that are carpeted. Since they now have an R-value of, say 3 above and nothing below, my concern would be that they would loose heat to the space below. That heat would make it's way into the second storey through the wood floors of other rooms, so the heat isn't "lost" just missing from the carpeted rooms. The estimates that I have suggest about double the heat-loss downwards for these carpeted rooms vs wood floor, but that's for "standard" construction, not a large empty plenum underneath.

Is it a valid concern? Is there an easier way to solve it? You're probably right that just a little insulation under the floors in those rooms (2) would be effective (and cheaper, easier, etc.). But fixing a problem after the fact would be problematic.

So the rooms on the second floor have different floor coverings(?) and it's all one zone, resulting in an imbalance? If it's micro-zoned by room it would still work if less than ideal. The air is fairly insulating, and the heat transfer laterally from the carpeted room to the adajacent hardwood-floored room via convection currents in the dead-space just isn't going to be huge. The bigger issue would be losses to the rooms below- you need to isolate them from one another so that the heat radiating & conducting heat from the high-emissivity sub floor above doesn't heat ceiling of the room below. Some amount of isolation is necessary whether micro-zoned by room or not. Normally this isolation would be done with cheap R11-R13 batts between the joists. R3-5 of iso is proably not enough isolation, especially (but not exclusively) in the rooms with R2-3 of carpet & padding above- you'd need at least 1.5" of iso to really do it.

A half a foot of even sloppy-applied blown insulation atop the old roof would pretty much do it. Even thin spots of only an inch would have a higher R value than a half-inch of iso. If there's any way to access the space well enough to get a good visual on it you can insulate it well enough for the isolation you need without having to fill it completely. (You don't exactly need R100 of isolation here, eh? :-) ) If it's possible to get fully in there you can probably install batts (or at a minimum, radiant barrier, as lousy a solution as that might be.) Blowing insulation approximately evenly can be pretty easy if you have the rooms stripped to the sub-floor by drilling an array of insulating holes in the subfloor. (You'd have to patch the insulation holes in the carpeted room, but not necessarily in a hardwood room. 20 gauge steel would do for spanning a 1-1/4" hole, or you could glue & trim dowel, etc.) It might be just as easy or easier to just cut yourself a temporary access hatch and install batts though, snugged right up to the subfloor for minimal lateral air movement, if there's enough room for a skinny person to crawl around atop the old roof.

The space has as couple of relatively large rooms (16x27 with 12' cathedral ceilings) and 4 relatively small bedrooms between, say 150-220sq.ft. each. The two smaller ones are carpeted. It's proposed that each room is a zone (there's a hallway and 2 bathrooms TBD).

We have access under the floor in most of the space. There's insulation on the old roof. A prior owner had 1.5" of polyiso sprayed over the tar/gravel. There's no real thermal connection between the subfloor and the downstairs ceiling wherever the roof is still in place, and that's everywhere except about 6" around the edge (and a few miscellaneous access holes.

It's possible to add more insulation in there, I could blow insulation onto the top of the drywall but I would wonder whether it's worthwhile. If the concern is radiant, then it won't add any value. If the concern is convective since top and bottom are conditioned there shouldn't be much of a temperature differential to cause convection - and it's only got a few spaces round the edge to convect from.

I would expect the warm floor radiating onto the old roof to just affect the response time since heat radiated to the old roof will eventually convect or radiate back to the floor. There is a fair amount of thermal mass there, a good few tons. So that may add an interesting twist to balancing the load, but may not make enough difference to care about. It does make me wonder about the usefulness of some of the zones, but I do hate to heat rooms that are not in use.

Estimated heat load for the whole structure is about 11 BTU/sqft/hr (45kBTU for about 4100sq.ft.), so working in my favor is that I'm only trying to get a max of about 1500BTU/hr through the carpet.

With half the heat from the upstairs radiant zones going into the smaller lower-loss dead-space it'll be much warmer in there than in the conditioned space above or below- probably not a delta-T of 50F, but it might be 20F.  (Not all conditioned space is created equal.)

If the 1.5" of spray iso (~ R9) on the old roof is in good shape it will be enough zone-isolation between floors, but if you can get in there to put R11 or R13 batting snug to the subfloor of the upper rooms that would be better for putting the heat where you actually want it.  That would make cozy-warm dead-space run cooler and guarantee maximal peformance under the carpeted rooms.  Insulate where ever it's accessible, and don't sweat it if there a few places too tight to get into. 

Adding insulation between the old roof & the ceiling below would be less worthwhile than batts directly under the subfloor- don't bother with that detail.

Posted By Dana1 on 14 Sep 2010 04:16 PM
With half the heat from the upstairs radiant zones going into the smaller lower-loss dead-space it'll be much warmer in there than in the conditioned space above or below- probably not a delta-T of 50F, but it might be 20F.  (Not all conditioned space is created equal.)

How about we just insulate under the bedroom floors and vent the inter-storey-space?

There is actually a twist which I didn't bring up in order not to further complicate the issue.  The south half of the inter-storey-space is the return plenum for a thermosiphon air heater, and the old roof is thermal mass for that heater.  (There's a slight conflict here since we could end up using the hydronic heating to heat the return air for this, but I think that will resolve itself by basically not running the hydronic heat when the air panels are open or similar).  Since the old roof is the thermal mass, I'd rather not insulate under that floor.  It's wood so I'm not as concerned as the carpet.

So the south end has some vents, and could have more/power.  Any reason not to just vent the north end? Vent it into the hallway, that always needs some heat day and night, or into the bathrooms, they could always use a little more.  Without something to move the air, the hot air may stagnate between the joists and so unless we do a long narrow vent this might not work all that well.  It's also a fairly haphazard way to balance heat loads.

On the other hand, it's a lot easier than insulating.

Insulate the whole floor- seriously!

Charging that mass with the radiant heat reduces the solar fraction, slowing down or reversing the thermosiphon- it'll cut into your solar-fraction.

Venting it to thermosiphon the heat out of the thermal mass into conditioned space would still be a good idea if it doesn't interfere with the thermosiphoning of the air panel. But that's in ADDITION to insulating the radiant. Even insulated that space will run pretty warm with radiant floors above and a conditioned-space ceiling below.

The thermosiphon section is here.  The idea is that the thermal mass, including the old roof stays warm after the panels are closed.  As I say, the old roof is about 3" thick, not as shown.  That floor also has/will have radiant.

Maybe it's optimist to expect the old roof to radiate/convect back through the floor and just letting it vent back out again would be a better plan, but I think that using that thermal mass to get an hour or so of heat after the sun goes down is sound. The old roof is probably about 2 1/2 tons of material, add about another 1 ton for the flooring and you start to get some real mass.

I think that if I want heat through the floor, I shouldn't insulate.  So the rest of the floor can be insulated, but I'm still not sure about this area.

If you don't insulate, when the radiant is on you'll be putting 75-80F+ air into the thermal air panel instead of 65-70F air raising the panel's operating temp, significantly lowering it's efficiency. If you put R19 batts under the radiant, the efficiency of the panel will be pretty close to what it always was, even while the radiant is running.

If you reverse the flow of the solar panel with a fan, THEN you'd be solar-heating the thermal mass to a temp higher than living space which might give you the extra hour you're thinking about. But as it is, with the mass in the return path, it's temp is approximately that of the lowest-temp part of conditioned space, and can't deliver much heat to the living space until the floor of the room is significantly colder than the mass (which means an even colder room above.) The way it's currently running the mass is only delivering a slightly slower cooling time. If the radiant has been running AT ALL with an uninsulated wall the mass buys you even less.