Hi there,

We design a 2 storey ICF house, 1800 sq.ft. on each level, located near Ottawa, Canada.
The slab on grade will have hydronic radiant heating. No basement.
The question is how much insulation to put under the slab? The designer suggested 2" rigid foam but this doesn't seem enough to me, so I told him to double it, to 4" rigid foam. I am looking for a trade-off between insulation cost and energy savings.
Based on my research on the internet, the standard size of insulation is 2" like the designer said, but for colder climates like the one we are living in they suggest you could put more.
I even saw a super-insulated house having 14" foam under the slab.
What do you think?

Thanks,
Beno

Beno,
Maybe by doing the calculations you'll get a better idea of what your 'Trade Offs' will provide.

Let's assume that your ground temperature is 50° and that you want the indoor temperature to be 70°. That gives you a Delta T of 20°. The formula is: 1/R-Value x Delta T x Area. So here are your Heat Losses:
1" Styrofoam(R-5) 1/5 x 20° x 1,800 sqft = 7,200 BTU/hr
2" Styrofoam(R-10) 1/10 x 20° x 1,800 sqft = 3,600 BTU/hr
4" Styrofoam(R-20) 1/20 x 20° x 1,800 sqft = 1,800 BTU/hr

That should give you an idea of what your choices are.

Good Luck!

Thanks for the examples. I wonder though if 4" (R20) is enough? The ICF walls offer (in my area) an effective R35. And it seems to me that the insulation under a radiant floor is even more important than the walls, because the floor is heated. This is also a passive solar house.
Is the 1,800 BTU/hr heat loss through slab a small loss? I will heat the water with an oil boiler with 97,000 - 174,000 BTU/hr, so I guess it will have enough power for the 1800 BTU/hr loss (and of course for the heat loss from windows, doors, walls and ceiling).
Also, I am not sure that the ground temperature is 50°, because I will not bring fill around the slab on grade perimeter, the house will look like a 2 storey without a basement.

Be careful out there! 'Effective' is really meaningless. Once the heat escapes the interior layer of foam, it's trying to heat the concrete core. That core is usually fed by earth temperatures. What is going to 'win'? My guess is the earth. Many interesting scenarios can then be discussed.


Well, maybe. My current house has no insulation below the radiant slab(last minute & budget issues). Am I losing efficiency? Absolutely! But what is happening is that the ground below the slab is being heated. So, by force, I'm creating a bit of more of a Heat Sink. That fact may keep the area warm a bit longer, and it's the most comfortable area in the house(upper floors don't have radiant).


Yes. But remember it's constant due to the fact that the earth temperature is constant.


Have a heat loss calculation performed. That boiler sounds to me to be too large.


You should insulate along the edges of the slab. That will help.

Good Luck!

Thanks again for your comments. I am insulating along the edges of the slab because I use ICF with footings on the bedrock. The slab will be between the ICF walls.
I will go with 4" foam all over under the slab, I have enough thermal mass in slab and I prefer not to warm the earth.

Is there a formula to help me decide the size of the boiler? Let's say that the total heat loss of the house is 10000 BTU/h. And the house has 5 baths for 6 people. Will a 100000 BTU/h boiler be too much?

Thanks again,
Beno

Yes! NRT.Rob might be able to give you a more exact answer, but most people I know size the boiler for the Heat Loss and maybe plus a little. 90k is a lot more than a little. I would think that for a 10k Heat Loss a boiler of 12k would do the job. Please remember that your Heat Loss is for the coldest temperature expected at the site. So, it's only running at that capacity a fraction of the year!

And, by all means, purchase a modulating condensing boiler. You'll save even more $$$.

Good Luck!

I use 2 Inches of foam with a product called slabshield under it all. And It all depends on how many square feet your heating for the boiler size + heat loss. Heat loss will not tell you what size boiler you need alone. Also you want to make sure its sized right or it will ether short cycle or run way to long. It looks like your heating 3600sq. ft. Id recomend an 80,000 BTU gas boiler thats at least 93% AFUE. You can get them up to 96% but there a bit more spendy..

Weil Mclain Ultra 80

One I use the most
Slantfin Bobcat (4:1 modulation provides input range of 30,000 to 120,000 Btuh)

A design day heat loss of merely 10kbtu/h for a 1800' 2 story in Ottawa would be on the extreme low side for a conventional house.  You can't just take a wild-assed-guess, do a ACCA Manual-J type heat loss calculation to figure the boiler size.  (That's the standard formula for boiler sizing, on which many heat loss software calculators are based.)  I expect your design-day temp is -20C or colder, and the boiler needs to support the heat load at the 97th percentile binned-hourly weather data for your area or you'll not keep up. But oversizing it results in higher numbers of burn cycles (even with modulating boilers), and lower efficiency overall.

If you're using the same boiler for domestic hot water, 5 showers running simultaneously will easily outstrip the full output of a 100K boiler unless you have considerable storage capacity.  If any of the showers are on the second story, DO look into drainwater heat recovery- it may allow you to downsize the boiler as well as the indirect-fired HW heater.  Even though the heating system uses the most fuel, your peak loads will clearly be HW.  You may need in excess of 100 gallons of storage if you're all taking showers at once, and even then you'll have to be quick about it.  If it's at most 2 showers at a time you can dial that way back.  With a drainwater heat exchanger you'll get ~25-30KBTU/hr back out of the heat exchanger during a shower minimizing the need for upsizing tanks or boilers, and you can size the storage for what it takes to fill the largest bathtub. (Drainwater heat recovery only works when the drain flows as the same time as the water. Bathtubs are fill first, drain later batch draws.)

Also, for estimating the heat loss out the bottom of a radiant slab you can't use the room air temp for the delta-T, you have to use the SLAB temp, which will of-necessity be above 70F to keep the room at 70F.  The exact temp will be a function of the total heat loss of the room, but 75F would be a more likely heating-season number to work from than 70F unless it's a truly super-insulated house.  Depending on soil type you may have real R value (and thermal mass) at the center slab to help you out, but at the slab edges (anything within 3/4m of the exterior wall) the heat loss will be considerably higher over the course of the winter, not mere ground temp. 

You may find this document useful:

http://www.toolbase.org/PDF/DesignGuides/revisedFPSFguide.pdf

You'll have to interpolate the maps to figure values for Ottawa, then add R5, since it's a warmed radiant floor, not just a heated space.

Just eyeballin' it using figure 4 and table 10, I guesstimated R10-15 for the whole floor, R20-25 along the perimeter if you add R5-ish for it being radiant slab.

"Heat loss will not tell you what size boiler you need alone."

Your kidding right?

A proper heat load analysis tells the experienced designer everything he needs to know!

A good radiant heat load program will include many factors you may not have considered e.g. the load will be lower than with conventional forced air or even fin-tube heating systems. It is all about temperature differential since the ground temperature is more stable the further down you go the less return on investment you will see as the insulation gets thicker.

At the perimeter it is a different story of course as the outdoor ambient will be much colder driving the differential temperature higher.

The simple way to get a handle on this is to look to the local frost line (or the depth your builder want to sink your foundation walls). It is this area where your attention and insulation should be focused.

This information has been around for a while so talking out of one's hat is unacceptable.

http://www.nrc-cnrc.gc.ca/obj/irc/doc/pubs/rp/rp33/rp33.pdf

5 showers? With all due respect, a properly sized ModCon high efficiency boiler with a companion water heater, sized to the load, is nearly impossible to beat where natural gas is available. Comfort, economy and outstanding domestic hot water performance.

I would like to add, that we still use the 1998 International One and Two Family Dwelling Code for frost protect shallow foundations (FPSF) which enlists the air freezing index (AFI)- not to be confused heat degree days which will be a smaller number. Here in Minneapolis about 50% lower on the air-freezing index than Ottawa, the minimum horizontal insulation (at the edge or corner of a heated structure) is R8.6 and two feet in. We use R10 2" XPS for most applications. Where conservation is of particular concern this may be doubled at the frost line (four feet here) with minimal cost and reasonable results. But the your certified radiant floor designer must have told you this. I tried to attache a picture but the gbt's cache is too small for it.

I have to disagree with morgan that a standard mod/con and indirect setup makes much sense on a 5 shower system with only a 10kBTU heat load.

Maybe, if you use a tank in tank indirect or turbomax, and enlist it as a buffer tank, mixing off of it for the radiant. Maybe. But with a 10kBTU/hr load for heating, a mod/con would probably be a waste of money.

This would be a fine application for something like the Phoenix Evolution though if natural gas makes sense. If you're stuck with oil, something like the bock "water heater".. a big one... might make sense.

Be careful in here. Panelcrafter's btu calculations are absolute horse crap because heat loss under the slab happens by conduction, at right angles, in a medium (dirt) that is a relatively poor conductor. (i.e. a concrete molecule must heat the dirt molecule directly below it, which must heat the dirt molecule directly below that, and so on.) Underground heat loss is so gradual that isn't unusual to get a thermal lag measured in months, which to say that the soil under the middle of the slab is at its coldest in April and warmest in October. The notion that your slab uninsulated could bleed 32000 btu/hr forever is ludicrous. Canada has a foundation energy modeling tool that can give you energy savings at various insulation levels. http://canmetenergy-canmetenergie.nrcan-rncan.gc.ca/eng/software_tools/basecalc.html If memory serves, you can load the soil characteristics and climate unique to Ottawa.
There are two reasons to go thicker anyway: the slab itself is close to or in contact with bedrock; cloud cover and short days result in so little passive solar heat that you need to conserve it in the mass it's warming.
But as Dana says, the slab perimeter is much more important than the middle. Pouring the slab inside the ICF foam -- no contact between slab and stem walls -- is 80 percent of the battle won, probably more if your footers rest on bedrock.

I believe that Panelcrafter is close to correct. First, there are fairly straightforward formulae for determining below grade temperatures. The "yearly average" temperature is the average air temperature, but the swings (over the year) vary based on the depth below the ground, the water content of the soil, the insulative value (R-value) of the soil composition, and things like that. Here in Carmel CA, at a distance of 2.2 meters below the ground, the average temperature (like the air temperature) is 57 degrees, and the annual variation is from about 55 to about 59, assuming 10% wet sand, etc. It is surely true that the basement moderates the swings all around, if not raising the average temperature, but the calculations are still sound.

The problem with Panelcrafters calculations are (as far as I can see) that:
(1) The soil itself will retain some of the heat (because the soil itself may or may not be a good conductor of heat) and will mediate some of the "annual" swing, and raise the average temperature some.
(2) The BTU loss from the floor should be measured off of the average temperature of the floor, which is likely to be more than the stated 70 degrees during the winter. This probably doesn't make that much of a difference, but it is the apprpriate number.
(3) The soil structure itself has some insulative value, so the appropriate R-values are not 5/10/20, but probably 7/12/22. There are some farily straighforward scientific experiements that you can do in a steady-state underground room that you can do to "measure" the R-value of the floor. Of course, this won't do you any good after the concrete is poured, but the methodology does exist. Sand, for instance, is moderately insulative, clay less so, and bedrock a lot less so. None of these materials will "lower" the R-value, but they may raise it a little or a lot.

Jeff

Those scientific experiments have already been done and incorporated in Basecalc, which offers weather and soil profiles for San Francisco by the way. On an earlier occasion, I used Basecalc to show Panelcrafter that 2 inches of XPS under a 1000 sf basement floor in Minneapolis lowered the season-worst basement heat loss from 5,700 btu/hr to 5300 btu/hr, or 400 btu/hr. Now a radiant slab on grade is different, but only by degree:  http://www.radiantpanelassociation.org/i4a/pages/index.cfm?pageid=420


Or you could listen to engineering professors who have spent decades devising models and testing them against real basements. Turns out Basecalc, the Canadian government's basement heat loss modeling software runs fine under W2000, and XP as well, except for some extra steps necessary to launch and exit the program. (Since upgraded) A couple of runs blows a basement-sized hole in the conventional wisdom on this forum.
I modeled a 1,000 SF basement with 8 inch poured walls, and a 4 inch poured floor five feet below grade (roughly translated from the metric measurements preferred by those pesky Canadians.) I conditioned the basement to a year-round 70 degrees. I modeled heat loss using the software's soil and weather profiles for Wash DC, Minneapolis and Denver. Minneapolis makes the strongest case for insulation, not surprisingly, but the others were in line, so we'll stick with the worst case in Minnesota.
With no insulation, our Minneapolis basement bled 61.5 gigajoules of energy a year. A GJ is roughly equivalent to an MCF of natural gas, so it is fairly easy to adjust the loss for furnace efficiency and translate it to dollars. That would be north of $1,000/yr in the case of no insulation in Minneapolis and natural gas at $15 per MCF.
In the second pass, I added two inches of EPS II to the basement's exterior walls in their entirety. The loss fell by more than half, to 28.8 GJ, for savings of $500+ year that are easily justified.
But that would be the end of the no brainers. Adding 2 inches of XPS under the slab reduced annual energy loss to 26.4 GJ from the 28.8 GJ above. So XPS at $1/SF saves $40/year, assuming $15/MCF gas burned in a high efficiency furnace. The payback is an emphatically unacceptable 25 years. It should be noted here that fuel oil at $5/gallon reduces the payback to 10 years, or just within reason. I am not saying don't insulate. I am saying think about it first, including how that $1,000 could be better spent.
Why XPS disappoints is an excellent argument for studying problems rather than throwing money at them. Butting wall and slab foam to the footers leaves thermal bridges at the precise points that soil can diffuse a fair amount of heat: along foundation edges and most particularly at corners.
In a fourth run, I moved the EPS to the interior of basement walls, put a thermal break (RSI=1) between the wall and the slab, and placed 2 inches of XPS under the outside edge of the slab, insulating its first 3.28 feet. The energy loss drops to 23.5 GJ, for a equivalency savings of 5.3 MCF/yr over exterior EPS alone, or $88/year.
In a final run, I extended the two-inch XPS in the model above to full coverage under the slab, vs about 40 percent coverage in the edges-only scheme. Full coverage reduced energy loss to 21.1 GJ, for savings equivalent to about 2.4 MCF/yr, or $40/yr once again. While the payback on the extra $600 of XPS is better at 15 years, it still isn't good. Clearly, basement floors are not major energy concerns, even in Minnesota.
Lest you think I am fudging this somehow, I refer you to a 2005 study by a Canadian goverment task force: "Performance Guidelines for Basement Envelope Systems and Materials" http://irc.nrc-cnrc.gc.ca/pubs/rr/rr199/rr199.pdf The 193 page study presents best practices and uses Basecalc to perform cost/benefit analyses. It ignores slab insulation except in a single word on page 16: "(Optional)"

Well not sure about your post. But I heat a 40X60 (total of 2400SF) home with just slab heat. For just around or under a $1,000 a year in Minnestoa. Using Natural gas. So not sure how I could be losing a 1,000 a year in the ground?

your slab is completely uninsulated on all sides, and below?

Minn NG prices are roughly half of the $15/mcf I used in my calculation, although DOE tables show they hit $18 briefly in the late, great energy boom. Also, there is a consistent thread in the literature that energy models like Basecalc tend to overestimate heat loss and consumption.

This is interesting, It runs counter to what is required in Washington state, and my preferred design (2 inch foam) But it brings the question to light that a 4 foot perimeter under slab of 2 inch and 2 inch board below frost line around the perimeter. But now we come back to this concrete barrier foil... I can here it already but the stuff is cheep, is about R-1 and is a vapor barrier at a modest cost,
so if one of the engineer number cruncher types on the forum could consider the calculation savings with R-1 insulation and Perimeter protection this would be interesting,

To Morgan, Rob, Dana and others Best wishes for the New Year, Peace on earth
Dan

Best wishes to you, but all concrete foil is, is an overpriced vapor barrier. R1 does so close to almost nothing it's not even worth considering it. it's a total and complete waste of money.