Posted By Dana1 on 29 Nov 2010 06:03 PM
The floor temp of the radiant heat depends only on the heat load it needs to deliver- at Vancouver's heat loads it'll be much more modest than in the midwest. The water temps depend on the system/tubing/plating type and the heat load per square meter of floor space. The source of that hot water isn't particularly relevent, other than the efficiency at which the source can deliver it. With geo it means you really want to run it below 45C most of the time, if you can, and I'm pretty sure that you can, even with the cheap thin plates. With condensing natural gas you need to run it under 50C if you want to keep it in the 90%+ efficiency range, but even at 60C it'll be 85% efficient. With either using an "outdoor reset" control scheme to run the temps at the lowest that keeps up with the load (by tracking outdoor temps) improves the seasonal efficiency. Most of the season you'd need no more than 35-40C water, but on the very coldest nights you might need 45C+.

We just came out of a cold snap in a climate very close to Vancouver's (just across the water on the Island).  We had no problem keeping the house warm with water flowing below 40C (104F) in our hydronic system.  It's reltively new construction insulated to code, but not much beyond code.

Posted By Dana1 on 29 Nov 2010 06:03 PM
/> docjenser: Neither of the studies you posted compared with/without reflective insulation (indeed, none used it), but both compared suspended-tube vs. plated systems. The primary difference in heat output at a given water temp between above the subfloor systems vs. staple-ups is the R-value of the subfloor, but the quality of the conductive bond with the PEX can trump that. If you look at figs 5.2.2 and 5.2.3 in Khanna's thesis you'll note that at 100F & 110F, with heavy extruded plates below the subfloor you achieve the same performance as WarmBoard, Quccktrack, and Thermalboard, above the subfloor systems using thinner plates (that AREN'T a hammer fit the way extruded plates are.) Conduction out of the tubing is at least as important as the conduction from plate to flooring. And even the "enhanced" suspended tube system UltraFin is nearly an order of magnitude behind them.

PEX stapled up without plates still outperformed suspended tube UltraFin by 2x at these temps. Given the relative stiffness of PEX (compared to EPDM), that isn't very much conductive contact area at all, yet it literally doubled the heat transfer rate! Were radiated heat transfer a large part of the equation that likely would not have been the case. UltraFin is an attempt to conduct more heat out of the PEX, but convect that heat to the sub-floor, and is barely half as effective as the most basic plateless staple-up.

Thin plates will deliver about ~75% of the performance of heavy plates, but that would still beat an unplated PEX staple-up by 50%. IIRC EPDM runs a bit behind the thin-plated PEX but better than an unplated PEX staple up. Too bad Khanna didn't test those as well.

I don't question that convection is a very efficient way to transfer heat, that is why a like the direct staple up against the subfloors, dislike the suspended systems, and I pointed out that the mentioned studies show a 70-100% better heat transfer of plates vs no plates. But the studies are flawed in the way that they only resemble the scenario of new builts, where the subfloor is pretty airtight, and the wood is a pretty bad conductor of heat.
So with an old floor, there are all the cracks and air gaps, were intrusion of heated air from a cavity below into the conditioned space becomes a bigger factor. That way the gap between plates and no plates closes, not fully, but significantly. In new construction, I would always use top of the floor applications, but in older houses, staple up is the way to go. So the question is whether the remaining benefit of plates is worth the extra cost in those applications. We are getting an average of 25 BTU/sqf at a 120 degrees F supply temperature with the onix product, resulting in about 80 degrees floor temperature, enough to heat our old houses here in Buffalo. It requires higher temperature than PEX but does not expand, so it does not bend down and stays in closer contact to the floor. Also the staples flatten it out slightly, increasing the contact area. The cavity helps with heat uniformity, we measure less than 4 degrees between the coldest and the warmest spot running with 120 degrees supply temperature. Combined with an outdoor reset, 120 degrees temperatures are only necessary in cold extremes. So this actually heats our old staple up houses with geo pretty well in our Buffalo NY winters, including my own. Like I said, I will put some WELs on a couple radiant systems in a month or so, so everyone can see by themselves.

docjenser,

You said that you would always use a top floor application for a new build. Our house is practically that. It was a full gut and we tore out the existing hardwood floors. There are a few areas where we still have old subfloor, but it is in good shape. Would you still recommend it? Does anyone have any experience with Uponor's Quiktrak system? Is it more or less responsive that Rehau's Raupanel?
I am leaning towards such a system for a few reasons: 1) it seems that we could operate at a lower temperature 2)I have heard of staple up systems creaking (although this was not with geo but with a boiler that runs at much higher temperatures. 3) staple up systems seems to be less responsive that top of the floor systems. Thanks

Convection is a LOUSY way to heat- did you mean CONDUCTION?

The most-important conductive interface isn't plate-to-floor, but rather plate-to-PEX. Plates work just fine on old ship-lap or T&G full of splits & knot holes as long as there's at least SOME amount of flat contact with the wood. Getting the heat out of the tubing is the bigger problem, which is why the tighter fit of extruded plates works so much better than sheet-metal versions. Contact with the wood is about the same with either.

I'm somewhat skeptical of the 25BTU/ft @120F number with EPDM staple-ups- extruded plates 8" o.c. on a perfectly flat plywood subfloor won't deliver that performance unless the combined R of the subfloor+ finish flooring is down around ~R0.25, which would have to be tile on a too-thin subfloor. A 1x ship-lap subfloor with hardwood runs about R1-R1.5, at which point extrusions deliver an honest 15-18 BTU/ft@120F. There's either a mis-calc or a mis-measurement going on if you're coming up with 25BTU/ft @ 120F with Onix. (Or mayhaps it's laid out 2" o.c. ? :-) )

How are you measuring it? About half that number seems possible without violating the laws of physics- more than that feels a bit like wishful thinking.

FrankieD: Be prepared for some sticker shock on the above-the-subfloor systems. They're more responsive, but they also have more apparent striping issues. Extruded plate systems are nearly as responsive (depends on the thickness of the subfloor & flooring), and don't exhibit any of the creaking issues associated with bang-bang controlled thin-plate systems and unplated staple-ups Thin plated PEX generally only creaks on cold-starts. If you use continous flow and outdoor reset you almost never hear them.

I would always prefer top of the floor applications, for exactly those reasons you have mentioned. Staple up are great in older houses, when you do not want to touch the existing hardwood, but when those are out and gutted, you can stat fresh with a new subfloor and I would use radiant above the floor. The last install we did with Quicktrack is about 1.5 years ago, and it was turning out great, running different zones with either 3/8 or 1/2' pipes, with 8 or 12" OC. Turned out great with geo, satiesfying a 50000 BTU heatload. Mean circulating fluid temp was 95 degrees F with a 20 degree F delta. Runs very efficient, gave me about 10 BTU/sqf under those numbers. Now it nicely heats the house, but I do not know the difference in heat uniformity.

Maybe I should be more explicit about the roots of my skepticism about the 25BTU/ft @ 120F number.

Assuming R1 of sub-floor + floor (it's probably closer to R1.25-R1.5 for ship-lap + hardwood) and a 70F room, a uniform temp of 120F below the subfloor would yield 50BTU/ft heat flux through the floor.

With Onix 4" on center you half 3 striped of contack 1/2" -3/4" wide- lets be generous and call it 3/4". 3 x 0.75 x 12" = 27 square inches or a bit less than 1/5 of a square foot, but let's be generous and call it a full 20%, which would give 50 x 0.2= 10BTU/foot.

Now the convection & radiation off the back side of the tube puts out a significant amount, but it's probably less than half the heat flux that you'd get with the conductive contact area, but let's say it's half. That adds another 5BTU/ft...

... for a grand total of 15 BTU/ft, with generous assumptions. Reality is probably on the order of 10-13 BTU/ft @ 120F average water temp.

To hit 25BTU/ft would require that the non-contact sides of the tubing was delivering 1.5x the heat that the contact side is. (Not very likely.), or the R value of the sub-floor + floor (+ tubing itself) adds up to less than R0.5 (also not likely.)

OTOH, if you're running 35F in the room rather than 70F...

It's an oversimplified model, but I believe I'm skewing the errors all in the direction of higher heat flux, not less. What have I missed?

Posted By Dana1 on 01 Dec 2010 10:10 AM
Maybe I should be more explicit about the roots of my skepticism about the 25BTU/ft @ 120F number.

Assuming R1 of sub-floor + floor (it's probably closer to R1.25-R1.5 for ship-lap + hardwood) and a 70F room, a uniform temp of 120F below the subfloor would yield 50BTU/ft heat flux through the floor.

With Onix 4" on center you half 3 striped of contack 1/2" -3/4" wide- lets be generous and call it 3/4". 3 x 0.75 x 12" = 27 square inches or a bit less than 1/5 of a square foot, but let's be generous and call it a full 20%, which would give 50 x 0.2= 10BTU/foot.

Now the convection & radiation off the back side of the tube puts out a significant amount, but it's probably less than half the heat flux that you'd get with the conductive contact area, but let's say it's half. That adds another 5BTU/ft...

... for a grand total of 15 BTU/ft, with generous assumptions. Reality is probably on the order of 10-13 BTU/ft @ 120F average water temp.

To hit 25BTU/ft would require that the non-contact sides of the tubing was delivering 1.5x the heat that the contact side is. (Not very likely.), or the R value of the sub-floor + floor (+ tubing itself) adds up to less than R0.5 (also not likely.)

OTOH, if you're running 35F in the room rather than 70F...

It's an oversimplified model, but I believe I'm skewing the errors all in the direction of higher heat flux, not less. What have I missed?

I think where your assumption most differs is the R value of the floor. I have attached a Onix Staple-Up Nomograph showing a supply temperature of 125° with a BTU load of 25 BTU/sqft and floor surface temperature of 81°F. The R value here is 0.6.
My own wood floor in my 1917 built house has 1/2" gaps between the subfloor planks, and quarter sawn oak sanded down a couple times. The oak is directly nailed onto the subfloor, no grooves here. My R-value is likely way below 0.5. Pretty much all the heat from the cavity below gets transfered upwards.
While the Nomograph assumes a temperature delta of 20 degrees (125 supply, 115 mean, 105 return) I work with a couple tricks to increase the efficiency when the output gets too close to the balance point. First an outdoor reset, so I ran my buffer tank at a lower temperature 95% of the time, being easy on the heatpump. Second I use a variable speed circulation pump which increases the flow rate on very cold days. That way my return temperature is up by about 10 degrees, and my mean temperature is up by about 5 degrees, resulting in about 3 BTU/sqf higher heat flux, all with the same supply temperature of 125 degrees F. That way, if my infrared can be trusted, I got 83 degrees surface temp out of the floor on a 5 degree F day last year. If I need more, I use PEX, which gives me about 4 BTU/sqf more flex, or put them closer together (6 inch OC) or use plates. You have to be creative and play with the cards you have. The plates are not great for the retrofit since the nails sometimes stick through the subfloor in those old houses.

Posted By Dana1 on 30 Nov 2010 06:18 PM
Convection is a LOUSY way to heat- did you mean CONDUCTION?

Yeah, sorry I mixed that up!

I'm not quite buying that 3/4" of subflooor (any species) plus 3/4" of hardwood (any species) comes in at anything less than R1 total.

Typical hardwoods would come in between R0.6-R0.7 by themselves,

A subfloor 3/4" fir or hemlock would be around R0.75. (Plywood at 3/4" is about R0.95 )

How does R0.6 + R0.7 add up to R0.6?

To get the R value of oak down to below R0.5 by itself it has to have been sanded down around 0.6" from the nominal 0.75" (credible, I s'pose), but the gappy-crappy planky stuff is still giving you another R0.6 unless it's SCARY thin.

Your floor + subfloor is probably more than twice R0.5, not "..way below R0.5."

See: http://www.warmyourfloor.com/pdfs/R-value%20Subfloor.pdf

Why anybody makes nomographs for R0.5 is beyond me, but I've seen them as low as R0.25 from plate manufacturers. Real-world flooring Rs start at about R1 for linoleum or tile-clad plywood and go up from there.

It occurs to me that steel subfloors with thinset & tile above might come in at around R0.25, so I s'pose it's not IMPOSSIBLE...

Guys,

What's consensus on the reflective barriers? I'm doing an installation inside joists using 5/8" tubing on 8" centers on my own place. The floors are mostly 120-year-old pine, and I was planning to use R-13 plastic-enclosed glass for acoustic purposes under a radiant barrier. (I got a deal on 2000 sq ft of glass rolls at a 60% discount.) It sounds like I could cut out the radiant barrier and save a few hundred dollars. Any reasons I should use the barrier + glass, or is this just a waste of money?

FYI, I'm in Pittsburgh. Design day is down to -5, but I'm in an interior row house with R-25 walls and R-45 ceilings. Heat source is a condensing natural gas commercial water heater - so no real temperature concerns. Design is for @35k Btu/hr out of 2500 sq ft of available heated floor (14 Btu/hr/sq ft).

Should've mentioned that I am absolutely planning to use plates, though. (Formed, not extruded, 2-channel plates so both tubes between each joist are in the same plate.)

Posted By Dana1 on 01 Dec 2010 05:58 PM
I'm not quite buying that 3/4" of subflooor (any species) plus 3/4" of hardwood (any species) comes in at anything less than R1 total.

Typical hardwoods would come in between R0.6-R0.7 by themselves,

A subfloor 3/4" fir or hemlock would be around R0.75. (Plywood at 3/4" is about R0.95 )

How does R0.6 + R0.7 add up to R0.6?

To get the R value of oak down to below R0.5 by itself it has to have been sanded down around 0.6" from the nominal 0.75" (credible, I s'pose), but the gappy-crappy planky stuff is still giving you another R0.6 unless it's SCARY thin.

Your floor + subfloor is probably more than twice R0.5, not "..way below R0.5."

See: http://www.warmyourfloor.com/pdfs/R-value%20Subfloor.pdf

Why anybody makes nomographs for R0.5 is beyond me, but I've seen them as low as R0.25 from plate manufacturers. Real-world flooring Rs start at about R1 for linoleum or tile-clad plywood and go up from there.

I keep saying, if heat has enough cracks and spaces, that changes everything. All the R1 values you mention are about conduction, that is all great, but in reality, heat travels upwards in many ways and forms. Lets say you have a pretty airtight floor of plywood, you have an R-value of 0.95. Imagine now you have a 1/30" gaps every 2.5 inches, where a different temperature of air can penetrate trough, how much do you think your R-value of your floor is now? I liked your analogy with the frying pan. So you can do the test right now. You have a nice airtight window with a high R-value. Now open this window up by 1/10 of an inch, how much did the R-value of that window as a whole change? As everyone knows, air exchange between 2 spaces with a high temperature delta changes everything, that way the R-value becomes quite insignificant. What you call "gappy-crappy planky stuff" I call real world staple up in older homes and a game changer for heat transfer, allowing for installment without plates. But I stand to my offer, I will put a WEL onto a radiant system with temp sensors on the water, in the cavities, on the floor and in the room. That way everyone can see. Just give me 1-2 months. As I said, it is kind of busy right now.

Posted By adamskj on 01 Dec 2010 06:28 PM
Guys,

What's consensus on the reflective barriers? I'm doing an installation inside joists using 5/8" tubing on 8" centers on my own place. The floors are mostly 120-year-old pine, and I was planning to use R-13 plastic-enclosed glass for acoustic purposes under a radiant barrier. (I got a deal on 2000 sq ft of glass rolls at a 60% discount.) It sounds like I could cut out the radiant barrier and save a few hundred dollars. Any reasons I should use the barrier + glass, or is this just a waste of money?

FYI, I'm in Pittsburgh. Design day is down to -5, but I'm in an interior row house with R-25 walls and R-45 ceilings. Heat source is a condensing natural gas commercial water heater - so no real temperature concerns. Design is for @35k Btu/hr out of 2500 sq ft of available heated floor (14 Btu/hr/sq ft).

If you only need 14 BTU/sqf, and you have a conventional boiler, and you are using plates anyway, and you use R-13 insulation underneath, I would say you have enough heat going up (and you can always add more), you can easily skip the foil. But this was not the scenario we were discussing. I would even say, you can skip the expensive plates. From the studies posted earlier in the thread, you get about 11 BTU/sqf with staple up without plates, and about 23 BTU/sqf with plates out of it, without reflection on the insulation, at 120 supply water temperature. So either way, you should be good to go.

Posted By docjenser on 01 Dec 2010 07:09 PM
  but in reality, heat travels upwards in many ways and forms.

Heat moves to cold. Up, down or sideways...heat always goes to cold. Hot air rises.

Bergy

Posted By Bergy on 01 Dec 2010 09:53 PM

Posted By docjenser on 01 Dec 2010 07:09 PM
  but in reality, heat travels upwards in many ways and forms.

Heat moves to cold. Up, down or sideways...heat always goes to cold. Hot air rises.

Bergy

OK...Heat in form of hot air raises

How about if you guys give doc some slack here.   He is reporting how his systems are working.  He shouldn't have to prove why it works.  If it works, it works.  It doesn't matter if you guys can't understand why it works.

He has offered to install monitors to give data to support what he is saying.   Let's give him the time to get it done.

Posted By geodean on 01 Dec 2010 11:21 PM
How about if you guys give doc some slack here.   He is reporting how his systems are working.  He shouldn't have to prove why it works.  If it works, it works.  It doesn't matter if you guys can't understand why it works.

He has offered to install monitors to give data to support what he is saying.   Let's give him the time to get it done.

As an engineer, the only thing that bothers me more than not being able to figure out why something doesn't work is not being able to figure out why something does work that shouldn't! I agree with your sentiment, though. Let's see what the data looks like when he has time to do it.

If it does work when engineering suggests it shouldn't, I (inwardly) attribute it to clean living and purity of heart, stroke out an invoice and hit the road...

Guys......I agree that no one can beat the physics, so there needs to be a good explanation for that. We will get to the bottom of it, and may be we actually learn something here, myself included. Keep in mind that the process of discovery usually starts with an observation which cannot be explained by the current thinking....