Thursday, December 18, 2008

Insulating the pool

Since this is my first posting about the pool, let me show you around a bit.  South is to the left.  You can see all along the south side of the pool there is a two-foot-wide shelf.  The water over that shelf will be 12 inches deep in the shallow end and 18 inches deep in the deep end.  In the middle of the south side you can see where the hot tub will go.  At the far end (west), you can see the shelf that will eventually support the automatic cover vault.  To the left of the pool, at the far end, you can see the pump vault that will hold the pumps below ground, behind baffles, which should make them completely silent.  The top of the wooden form will be about three inches below the coping around the pool, so you can see that the coping is about 18 inches above grade on the near side.  That coping will be about 16 inches wide, and will form a sort of bench seat most of the way around the pool.  At the far end, the grade level will be raised to match the coping height, which because of the slope of the back yard will only be about six inches.

The blue stuff you are looking at is the 2 inch thick Dow Styrofoam Highload 40 insulation, most of which is glued (yes glued, with polyurethane foam) to the soil behind it.  On the bottom of the pool, the insulation sits on 4 inches of crushed drain rock, which sits on a geotextile fabric membrane.  The next thing to go into the pit is the plumbing and rebar grid, and after that we shoot gunite.

Heat losses for most pools are dominated by evaporation. We're going to be getting a safety cover, and one of the side effects of these covers is that they are close to vapor-tight. We'll keep it closed for most of the day and only open it to go swimming. As a result, I'm expecting evaporative losses to be quite small -- perhaps a half inch a month or so.  On our 18x46 foot pool, that's 260 gallons/month. Multiply by 2270 kJ/kg (water's heat of vaporization), and that's 70,000 BTU/day.  [Update: this evaporation estimate was dead on.]

Direct heat loss through the cover will be the largest remaining heat loss.  The 24-hour average outside temperature in spring is around 60 degrees, and the safety cover will be something like R-1, so loss from a 85 degree pool would be 520,000 BTU/day.  During the early spring and late fall, we'll probably lower this loss by putting a bubble-type cover over the safety cover.  (I wasn't able to find an automatic safety cover which insulates.)  The combination of the two covers will be around R-3, so losses will be around 180,000 BTU/day.

Most contractors tell me that the dirt under the pool is a fine insulator, but I think they really don't know what they're talking about. Houses use under-slab insulation to insulate their 70 degree interiors from the 55 degree earth heat sink. The pool will be at 85 degrees, which is twice that temperature gradient, so I think insulation will matter even more for the pool. In particular, I'm most concerned about the water table contacting the bottom of the pool in the spring and sucking all the pool's heat into an underground plume of warm water headed towards the San Francisco Bay.  [Update: the water table never gets to the bottom of the pool, as our dirt is well drained.]

Let's suppose that concrete conducts 1.7 W/m-K. To convert that to more familiar units, 8 inches of concrete would be R-0.68 (which is terrible -- single-paned windows are better). I'll guess that the dirt, even when wet, insulates a bit as well, so that the bottom of the pool is about R-2. My pool will have an average depth of 7 feet (it has a 10.5 foot deep diving area), so it'll have an exposed area of about 1790 feet. If the pool temp is 85 degrees, and the ground temp is 60 degrees, that's 537,000 BTU/day, about three times my expected loss through the top in the spring. That's why I'm insulating the pool.

The blue Styrofoam has an R value of R-10. It is designed to deflect 5% at 40 psi, and is rated for continuous (dead) load of 13.3 psi. 10.5 feet of water plus 8 inches of concrete will be 5.25 psi, well within the dead load rating of the Styrofoam. The stuff should compress 0.33 mm when the pool is filled, which I don't think will cause cracking anywhere (thermal expansion is probably more than that).

The insulation changes the bottom of the pool from R-2 to R-12, so now I expect to lose 90,000 BTU/day, which is a savings of 447,000 BTU/day.  To put that in perspective, on an average day my solar panels will deliver around 40,000 BTU each.  I think I can squeeze 12 of them up onto the roof.  The insulation is saving about as much heat flow as the panels put in!

To attach the insulation to bare dirt on the walls and shelves, I glued them on with Tap Plastics X-30 two-part expanding polyurethane foam, which I sprayed on with a pair of Wagner power sprayers.  The trick here was to hold the panels firmly (with a few hundred pounds of force) against the dirt while the polyurethane expanded, which takes about 30 minutes.  Once we did that the panels really stuck on well.

Note: I wore complete face covering, goggles, and breathed through an activated charcoal filter to take out the volatile organics.  It was impossible to get the crew to wear the same protective gear.  In the end, everything that was exposed got coated in a fine mist of polyurethane.  The warning label says skin will develop an allergy to the stuff with repeated exposure.  If you try this yourself, be careful, and be careful with your crew!

The soil at the bottom of the pool is firm clay, covered with filter fabric, covered with drain rock.  Initially I tried gluing the insulation to the drain rock, but this doesn't work well.  The insulation does not sit flat against the drain rock, and it rocks around a bit.  Worse still, the polyurethane is quite springy, and was deflecting about a quarter inch under my body weight.  This made me realize that the combinations of shelves, drain rock, and insulation is very bad, because the insulation on drain rock might settle, which would cause the pool to hang from the shelves, which would cause the gunite shell to crack.  If I had it to do over again, I would eliminate the soil shelves, which are only there to reduce the cost of the gunite by $1500 or so.

Instead, I ripped up all the bottom insulation that the crews had installed and reinstalled it myself.  I had already compacted the soil, first with the Bobcat, then with my feet (the crew were actually laughing at me), then with a heavy roller.  Then I compacted the drain rock with the heavy roller.  Then I cut the insulation into 15" x 24" squares and hand placed each one, tweaking the rock placement and vibrating the panel so that the rocks settled into a configuration which was flat for each panel.  That took about four partial days, and only got finished because Martha got into the pit with me.  I read later that a skilled laborer with one assistant is expected to place 100 square feet of under-slab insulation per hour, so I was about two times slower than that.

I'm still paranoid about settling, so I'm going to have the south and west walls of the pool done with #5 40 ksi vertical rebar on 6 inch centers.  #5 is strong enough to actually support the entire pool weight (450,000 pounds filled) on the soil shelves, should the shelves not creep under that load.  I also put a 1 foot chamfer on the edge of the shelf under the cover vault.  The fillet of gunite that fills that chamfer, along with the rebar, will spread the torque from the pool wall hanging on that shelf.

[Update: I ended up with a mix of #3 40 ksi and #4 60 ksi rebar, with about half the strength necessary to hold up the whole pool. I figured I only had to hold up one end, and the rebar guy objected strongly to #5 -- he bent most of the rebar with his right knee!]

Another issue is the bond strength between the XPS insulation and the gunite.  I've since built the pump vault, which is insulated as well, and the XPS/concrete bond there is perhaps only 1 to 3 pounds per square inch shear strength.  If I assume 1 psi for the pool shell, then it will take about 140,000 pounds of shear, which is substantially less than the 450,000 pound filled weight of the pool.  This shear strength is comparable to the shear strength of the polyurethane bond to the soil.  This pool shell will be sitting on its bottom.  The polyurethane is substantially more springy than the polystyrene, and I think it will deflect the 0.3mm in shear that will happen when we fill the pool.

I did not predict the cost well.  I used 130 2 foot by 8 foot panels, with a small amount left over that I am using to insulate the pool piping.  These cost $2800.  I used 30 gallons of the X-30, which cost me about $1500.  It took me and a crew of three guys four days to glue on the side panels, and I think that labor cost me $2500.  The bottom I did myself, but the labor for that would probably have been another $1000 or so.

You can try to compare that cost to the cost of the panel array.  The problem with the comparison is that the panels deliver the most heat when its warm, when you don't need it.  The second problem is that the panel array is limited by the size of my roof, so I can't just spend an arbitrarily large sum on panels.

That said, standard FAFCO-type panels are something like $300 each, and produce 25,000 BTU/day.  To generate the same amount of heat saved, you'd need 18 panels, which would cost $5400.  However, the panels don't really work at all in November and April, when you need the heat most.  The panels I'm using are SunEarth EP40s, which are glazed and insulated copper panels, and cost around $1100 each.  These will deliver around 35,000 BTU/day, and will work in November and April.  But I'd need 13 to match the insulation's savings.  And that would cost $14,000!  So I'm pretty confident the insulation will be a win.

I'm not quite done with the insulation.  You can see a pile of pink insulation at the southeast corner of the pool.  This is standard XPS, and it will be going around all the pool piping.  I tried to find XPS pipe insulation, but the stuff is hideously expensive because it is all custom cut from large billets, usually for municipalities who are insulating their sewer pipes against frost.  So instead I'm gluing together a little box of 2" XPS around all my pipe runs.  Because the pool pump will be running 24 hours a day, the heat losses from the pipes can actually be comparable to the pool shell itself if not insulated.  If you are building a pool, you might want to consider insulating your pipe runs even if you don't insulate the shell, because insulated pipe runs don't suffer any of the structural and installation issues that come with the shell.

Side note: I had a hell of a time finding a Highload 40 distributor. I ended up calling Dow directly, who referred me to White Cap Insulation in San Francisco, who sold me the stuff for $21.58 per 2" x 2' x 8' sheet. Note that standard 20 psi extruded polystyrene has a dead load rating of over 6 psi, so it would have been technically okay for my 5.25 psi of dead load. If I wanted to save around $600, I would probably have gone for the thinner stuff. It would also be easier to buy. I'm hoping that the extra-sturdy foam is buying me a little margin, which is nice to have because I don't have examples of other pools that have been built this way.  [Update: somehow I miscalculated the amount of gunite in the pool, and it's a bit larger than my estimate here.  So, I'm glad I have lots of extra margin on the load carrying capacity of the insulation.]

Side note two: the space down at the bottom of the deep end experiences really wild temperature swings right now.  Because it is down low, it is radiatively coupled to the sky and not much else.  When the sun is shining down there, it can get well over 100 degrees.  As soon as the sun goes down, the temperature plummets, and because the pool is a depression, chilled air tends to stay down there.

Side note three: If you read this post and insulate your pool, or know of a pool that's been insulated, you might leave a comment so that there is some repository of success stories on the internet for this idea. And please please post if there has been a problem with an insulated pool.


  1. nice application. Some more uses of geotextiles are here.

  2. Any post completion notes or pictures?

  3. We're shooting gunite on Wednesday (April 29). So we're not yet complete. The holdup has been my fountain. See my later posts for details.

  4. Great project. I would be interested in discussing with you some of your thermal analysis. Give me a call, Neil O. Anderson, 209-367-3701

  5. How about an update? I would like to know how construction went.


  7. good way to insulate your pool i am really going to work on this .. steps cool

  8. thanks for sharing great project with all. it is very helpful to those who want to insulate pool

  9. Great job done. Thanks for the information. Polyurethane foam roofing will also be found useful in this.

  10. Yes, polyurethane foam roofing is definitely another way to accomplish the same thing. Note that builders don't use that under floor slabs. I'd do a bunch of research on polyurethane interacting with the groundwater and bugs, etc, before I committed to that.

  11. This blog is full of useful information on pool. Thanks for the tips.

  12. I am building a in ground, outdoor pool in New Delhi, India using poured in place ready mix concrete and single cage rebar. 9 inch thick concrete floor and 6 inch thick concrete on walls using standard wood shuttering forms. Pool is rectangle L=60 Ft W=20 Ft D= 4 Ft sloping to 4-1/2 Ft. Concrete shell to be cured with water spray for 21 days and then waterproofed with thin coat 0.04 inch Laticrete waterproof compound (USA make) and tiled with locally made 1 inch X 1 inch glass tiles on mesh backing fixed Laticrete epoxy grout. Pool cover is Aquamatic, in wall rail, with hydraulic motor.

    Propose to insulate under floor and behind walls with locally made EPS (3 lbs/cu ft, compressive strength 50 psi at 10% compression) 4 inch thick on floor and 3 inch EPS on walls. Compacted crushed rock and geo textile as floor base. Will insulate piping with 2 inch XPS foam 2 lbs/cu ft density, pipe sections.

    Radiant heating of pool using hot water running thru PEX tubing 0.75 inch I.D. at 12 inch centers on floor and walls tied to rebar cage. Heat source is diesel fired burner and hot water boiler make Riello ( 150,000 BTU/Hr). Water in PEX tubing circulated using Grundofos pump 1 HP. Bubble type solar blanket may be used under pool cover if heat loss excesive. Want to raise pool temperature to 88 F in 5 to 7 days initially and then maintain continuously from end Nov to early April. Coldest period Jan and Feb ambient min 40 F and avg 50 F. Pool cover will be kept closed entire heating season except 2 hours per day when pool in use. Heating not required mid April to end November and pool cover can be kept retracted during this time.

    Spa is square 8 Ft x 8 Ft and 3-1/2 ft deep, adjacent to pool on the shallow side. Similarly insulated as pool but without PEX tubing. Spa to be heated using same hot water boiler as used for pool but using a plate exchanger mounted on the bolier skid.

    Request comments especially on possible problems caused by EPS compressing and cracking concrte.

    1. PEX is banned for use in pools in the US because it can't stand hi levels of chlorine!

    2. PEX would not be in direct contact with highly-chlorinated water. It would be embedded in the walls and have its own water or other fluid system.

  13. Brij,

    - If the local dirt isn't sucking water out of the concrete shell, you might think about putting plastic over the concrete. This prevents drying out perhaps better than water spray, since it protects the entire thickness of concrete evenly. If you spray (which is what I did) the concrete surface is going to cycle through dry spells and not cure as well.
    - Think about putting your pipework inside the concrete shell. In particular, the top 18" rim of your wall is probably a foot thick or so. Right below that, outboard of your structural shell but inside your insulating layer, you've got plenty of space to put your pipework. This eliminates the extra heat losses through the pipe insulation. (I lose more heat through my pipe insulation than through the pool to ground envelope!) It's also simpler -- all that pipe insulation is a hassle. Finally, you eliminate many of your PVC lines going from ground to concrete, which is a prime failure location.
    - Your pool is fairly shallow. Ground loading should be about 2 psi, which will compress the EPS 4% or 0.16 inches. That's probably okay, but you don't need all that insulation on the bottom. Why don't you knock down the bottom EPS to 2 inches thick?
    - I would think about temperature cycling all that concrete when you turn that diesel (EGAD!) boiler on and off. This might be your biggest risk, and I would not take it without a careful understanding of what you are getting into. Most concrete structures the size of pool shells have expansion joints, but your shell probably doesn't.
    - I find it unlikely you will get a bubble cover UNDER the Aquamatic cover. I use a bubble cover over my Aquamatic, which I think raises the R-value to about 1.5. Thus the top cover completely dominates my heat losses now that the ground insulation is well under control. Make sure that cover stays dry (pump it off right after any rain), or you will lose a lot more heat through it. If you've got a R-1.5 cover, ambient avg 50 F, and 88 F pool, you will lose 730,000 BTU/day through that cover. Your heater will do the job, except maybe in rain and wind, which will increase your losses to at least 2,000,000 BTU/day.

    Sounds neat!

    1. Ian,
      Thank you for your very helpful comments.
      Covering concrete with thin PE film is excellent suggestion.
      On thinking thru, I agree, putting the bubble cover under pool cover is not possible.
      I did ROI on 2, 3 and 4 EP on floor and find that 3 inch is best. Is your suggestion to bring down to 2 based on mm compression?
      I am not clear on how to fit the piping inside the concrete shell. If you send me an email address I can send you PDF of the pool drawings including the piping layout and the shell for your suggestion on how to fit the piping within the concrete.
      The Delta T for PEX closed circuit water entering and exiting the boiler is 15 F so the RCC shell will never see a temp. higher than 100F though due to heat transfer, I guestimate max RCC temp will be 95F. Do you think there may still be issues of expansion joint/cracking due to thermal cycling. The initial time to get from say 60 F to 90F will be 7 days.
      Again, thank you for your comments and I appreciate any reply that you may have.

    2. In my last post the temperatures were incorrect
      Temp of water in PEX entering circuit (supply)about 145 F
      Temp of water in PEX exiting circuit ( return) about 125 F
      Max temp of concrete about 115 F

    3. Brij,
      My suggestion to bring the bottom insulation down to 2 inches was based on the compression amount. If the cost difference in supplying make-up heat between 2 and 3 inches of BOTTOM insulation is not large, I'd go with 2 inches, as you'll have less compression. Erring on the side of less compression seems like a good thing, but that's a gut feeling. I cannot think of a specific way 0.16 inches of compression would actually cause a failure, aside from perhaps misaligning the deck to the coping a bit.

      Email address is my first name @ my last name dot com.

      Here's how I'd think about temperature stress on the pool shell:
      - first, do a quick calculation for how much the pool shell expands and contracts between high and low expected temperature of the pool water. Pools survive this all the time just fine, so it gives you a sense of the reasonable strains.
      - second, because you have a layer of insulation completely surrounding the shell, you actually know what the stress/strain ratio is there. So you can directly calculate the compressive stress on the pool shell from heating up. My guess is it won't be large.

      Here's the last calc: where the water from the boiler enters the pool shell, you have a local hot spot at 145 F (63 C). The water on the other side of the shell can be as low as 60 F, but actually it can probably get down to 50 F (10 C) if you've had to leave the pool unheated for a while for some reason. So you've got 53 C delta-T across, say, 100 mm of concrete.

      The hotter exterior of the shell is going to expand 10 ppm/C, or 530 ppm, relative to the interior of the shell, in that region. If I assume that the shell doesn't bend as a result, we can multiple by Young's modulus for concrete (30 GPa) and see that the interior and exterior will experience something like +/- 8 MPa of stress.

      The usual compressive strength of concrete is something like 30 MPa, so you've got 4x margin there, but the tensile strength of concrete isn't nearly so good (maybe 3-5 MPa). Given these numbers I would revisit your boiling temperature numbers. When the pool is at 50 F, and you turn on those boilers, the water returning from the pool to the boilers might be significantly colder than 125 F. Check that. Also, my assumption that the concrete is the same temperature as the pool water is probably off by a bit. If the temperature delta across the concrete at the hot spot is a lot smaller, you might have enough margin on the tensile strength of the concrete. But it needs to be a lot smaller, otherwise I suspect you might get a bit of delamination in the shell near your hot spots.

    4. Brij,

      The other thing that might work in your favor here is the insulative value of the PEX.

      I'd be a little worried about those glass tiles, specifically because all my glass tiles have shattered. Here's a test I did that fails for the tiles that failed in my pool, and passes for other kinds of tiles:

      Put a pan on the stove, fill with water, heat to a simmer. Fill another pan with ice water.

      Place tile samples in ice pan for 1 minute, then simmering pan for 1 minute, then back to ice pan, etc.

      100% of my pool tile samples failed on the first cycle. I found others where 20% of the sample failed in the first 7 cycles. Porcelain tile samples did not fail at all in 20 cycles.

      You might want to test your tiles before you put them in the pool. I wish I had.

  14. Hi Iain,

    Great job on your pool build. I'm in the process of building an inground pool out of ICF up in Canada. I just have a question about how you insulated your inground plumbing. I believe you just built boxes out of your existing foam panels around each pipe and then backfilled over top of the plumbing. Do you worry about water getting into the 'box' you created? I was thinking of back filling around the plumbing using pearock so water drains past it.

    Thanks for your thoughts...

    1. Copper cup,

      Yes, I built boxes with the 2" insulation. Those boxes are filled with the pipes, and the voids are packed with sand. I'm sure the sand is wet, and I'm not the least bit concerned about the dampness increasing heat conduction. A separate issue would be water pouring through the box.

      The boxes are made of, typically, a 9 inch wide bottom plate, two 3.5" tall by 2" thick side strips, and a 9 inch wide top plate, all surrounding a 3" diameter pipe. The top of bottom plates obviously overlap the side strips significantly. I'm not too worried about significant water flow from outside the box through the box, since the flow would have to go sideways.

      But don't mess with this idea, since the Australians have a better system: keep the plumbing inside the insulated pool envelope, by routing it just under the 1 foot x 1 foot bond beam around the perimeter of the pool. This requires less insulation, is therefore cheaper, has lower losses, and has far fewer opportunities for pipes protruding from the gunite to break.

    2. my pool equipment and gas fired heater is going to be about 50 ft away from the pool. what can be done to insulate the return lines on that length of a run? do you feel that all of the suction and return lines need to be insulated? do you think i should build a box around all of the lines that make the run from the pool to the equipment with the rigid foam that you are talking about?

    3. Justice,

      You'll lose nearly as much heat from the suction lines as the return lines: Your pool water will be at, what, 80 F? The water returning from a 80% efficient 400,000 BTU/hour heater at 40 gallons/minute will be at maybe 98 F. The ground is at 55 degrees F. So the suction line delta-T is 25 F and the return line delta-T is 43 F. My point is you are losing quite a bit from your suction lines too. Also, you lose heat from all these lines any time the pump is running, which should be all the time, and not just when running the heater.

      The main problem with building a box around the pipes is building any box at all. Once you've done that, putting all your lines in the box is hardly any more difficult than putting just the return lines in.

      Bottom line: box all your lines between the equipment pad and the pool. Use the rigid foam as I described, unless you figure out something better, in which case please post that here.

      I boxed a run of about 100 feet to my solar panels, along with about 200 feet of return lines around the pool. The run to the solar panels is fine, but in retrospect I wish I'd placed those return lines inside the insulated pool envelope underneath the bond beam, as I've said elsewhere.

      I put my pumps and as much plumbing as possible in an insulated below-grade vault. The filter and heater are above grade. I like the dead quiet and clean look, but clearing leaves from the filter lint basket is annoying.

    4. check out dritherm here:

  15. Wanted to get your take on using the insulation on a Fiberglass pool.

    My plan would be to insulate the ground as much as possible before dropping in a preformed Fiberglass pool. I was thinking the sprayfoam would be great but wanted to hear your thoughts.


  16. I don't know a lot about fiberglass pools. I'm pretty sure that most spas are fiberglass shells with spray-on polyurethane foam, so you've got a good set of reference cases there.

    The main issue is the huge weight of the water, and how that weight is going to be transferred to the ground. Polyurethane foam sprayed onto the fiberglass shell can probably hold the weight no problem. Ideally, get a specification for it's compressibility, or test a sample yourself. I'm not kidding - it's not difficult at all to test and it's important to do this stuff carefully.

    You need to make sure that however you spray this foam, you get a consistent high density result. You do not want something that is springy, because the deep end will settle differently than the shallow end, and that will stress the shell. I used highload-40 which deflects 5% at 40 psi, and I loaded it to 5.5 psi with the pool full.

    I suggest you shoot some test PU foam 2 inches thick, and set a wood block into the top before it cures. Cut the foam to the outline of the block to prevent spreading strain through the block. Then put a heavy weight on the block, and measure deflection. Note that for an e.g. 3.5 x 3.5 inch block, you'd want to put something like 250 pounds on it, and measure 1 mm of deflection. That's really stiff!