There was a thread here I read within the past week or so wherein one of the replies talked about insulating only around the perimeter of a slab to the footings.  This was for an unheated space (ie. garage); and with the rationale that heat from the ground below would keep the interior of the building warmer... or actually it did, by the poster's report.  I have seen that thought process, and several others.  Some say perimeter insulation only for unheated space for the above reason, some say at least R5 for unheated and at least R10 for heated, and some say as much as you can afford regardless.

I had planned to build a ~1200 ft^2 garage onto the front of my house, which would also have space for a workshop and 14' ceilings to allow for an automotive lift.  I wasn't going to heat it 24/7; but was going to have a propane tube heater to get it up to 50-55F.  With that in mind, I bought enough 2" thick Type IX borate-infused EPS to do the under the whole slab and the perimeter down to the footing.

Now, because life throws expensive surprised at you... sometimes 3 in one year... the garage has been downgraded to a more typical 26'x27' with normal ceilings and no shop space.  This will strictly be a place to park two cars, unheated; but I would like for it to be decently insulated.  Got some -42F wind chill days this year and that's hard on a vehicle.  I intend to add 2" of rigid foam to the outside of the structure.  Since it's no longer going to be heated, if I just do the perimeter down to the footings inside and outside of the block and not under the slab, I'll have enough 2" EPS left over to do all or most of the walls.  If I still do under the slab, I'll need to buy additional for the walls... and it will be XPS because the current EPS cost me a fortune.  No distributors in my area; so the shipping alone added nearly $10 to each 4'x8' sheet.  I'm not incuring the extra expense again just for 10-12 more sheets considering I'm already lower than expected on funds.

So what say ye experts?  Worth the extra expense to insulate under the slab for the above scenario; or can I just do inside/outside the block down to the footings with well-insulated walls and still be happy on those cold winter days?  It'll be an attached structure; so I figure some heat will get into the garage through the one shared wall.

If it will truly NOT be heated, you do NOT need any insulation anywhere. The indoor temp will become the same as the outdoor temp no matter how much you insulate it.

*** onto the front of the house*** so there will be heat loss from the house.

You can always add insulation to the walls but once the concrete is down, I think it will stay down.

When funds become available, I bet you will heat it.

Right, the correct decision depends on whether the garage will be heated…either now or in the future. Any heat loss from the house into the garage will result in some heating of the garage. If the outside air temp is less than the ground temp below the slab, not insulating the slab will result in some heating of the garage.

If you think that you will heat the garage in future, you should use slab insulation. Maximum benefit would be placing insulation around the slab perimeter and below the slab in the area adjacent to the perimeter.

I have been interested in observing the temperature in my unheated, insulated, detached garage/workshop lately. The garage is not insulated from the ground, but has 2x6 walls with R-20 cellulose in the walls, and I think R-39 loose cellulose in the ceiling. (The shop has a heater, but it has not been on this winter.) Ground water temperature is 53 F. The small (12' x 22') workshop on the south end of the garage has 2 low-e, high solar gain windows that are probably 40" x 60".

I assumed that the garage temperature floated up and down with the outside temperature, with some lag and some moderation based on the insulation. However, it stays much warmer than that pattern would suggest. For example, last night the low (air temperature) was -2 F, and the garage was the coldest that I have seen this winter at 39 F. I live in an area of high solar insolation so there is some solar heating, but I am guessing that the ground must also be adding heat into the building in the winter. I cannot say anything definitively, but I think that having no insulation under the slab must be keeping the garage from freezing. Of course, it might also make it harder to heat to a relatively high temperature. For my application, perimeter insulation might be handy, but probably not complete slab insulation under the garage portion.

I should add that a typical year here in the Colorado mountains includes about 7000 heating degree (F) days.

There is some missing info here.
Does the house sit on a basement or is it slab on grade? If basement (frost protected) then the garage will have to be the same.
Thinking on Lee's observations, if you are putting it on a stem wall, maybe 4' of insulation running OUT from the wall as in frost protection (sorry Jonr, could't wait for you to chime in) and up the outside of the wall. Might combine with the house heat loss to keep it above freezing. If your soils are dry enough, you might not need the underslab foam.

Where are you located?

I believe the answer to your bolded question is YES, but the under laying question that will likely drive your decision is how much heat? You may end up with a situation similar to what Lee described, but this all depends on how much heat gets stored/retained under your section of presumably slab-on-grade. I am still thinking that only insulating the perimeter and adjacent area might be your best solution. Not insulating under the central area of slab would still allow any availabale ground heat to move up into your garage.  You may also want to consider using a baseboard heater that only goes on when the temp approaches freezing, e.g., similar to what we use in well houses to ensure our pipes and filter housings don't get damaged.

Strawmyers, that may have been my post on that other thread you referred to. To reinforce what Lee and sailawayrb said, yes, insulate the walls and lid, insulate the frost walls/foundation of the structure down to the footing all around (I have just R10, and just on the inside of the wall), but don't insulate the slab. If the ground is at say 50 F and your inside garage temp stays in the upper 30s (likely), that gives you a decent 10 degrees or more to drive heat from the ground into the garage. You don't want to put insulation between your garage and its only heat source (other than the hopefully very small loss from the house), but you do want to insulate that mass of relatively warm soil under the slab from the cold frozen ground outside the foundation. You ought to consider a good insulated garage door, as that area will be your single biggest heat loss from the garage. Cut your foundation insulation to just R10 and put the difference into the garage door.

When I got up this morning it was -2 F outside. I just checked the thermometer (about a foot above the floor, well away from the door), and it read 38. What will happen, based on what I've seen through the winter (first with a thermometer out there), is that when the temperature outside warms up to the mid teens or so the thermometer will climb back to 40-42. About six feet in from the door there is liquid water on the slab, underneath the car, from the snow the tires tracked in yesterday and which melted once in the garage.

Posted By DickRussell on 07 Feb 2014 10:24 AM 
You ought to consider a good insulated garage door, as that area will be your single biggest heat loss from the garage.

Plan is for a Haas 2000 series door: 2" thick, polyeurathane filled, claimed R-17.66.  It'll either being a single 18x8 door or two 10x8.    http://www.haasdoor.com/residential/2000/index.html

Thank you for sharing your experience!

R-17.66 "Claimed"? Does this doormaker have any data on what the ACTUAL tested R-value is?

The issue with the garage door is not so much the stated R value as the air tightness of the door & gaskets. The R value is based on the thickness of foam within the door, but typically the doors are not thermally broken which diminishes the insulation value considerably, plus they are poorly gasketed. Consider that on a typical 9x7 4 panel door you have 60 LF of joints, all of which leak air. Keeping the interior of the garage above freezing will work best if you can keep cold air out.

If you had a sealed 24x24x8 box, insulated to frost on the perimeter, the tendency would be for the interior to stay at ground temperature (with allowances for the thickness of the insulation). But if instead you have a large box with lots of holes where cold air gets in, the interior temperature will drop. So seal the door well; perhaps adding some aftermarket EDPM gaskets. (conservationtechnology.com)

Since they're in the business of selling doors, not insulation they're allowed to be very "creative" about the thermal performance of the doors in ways that purveyors of insulation would run afoul of FTC regulation. Even the center-panel R is being exaggerated beyond what it's average performance would be. Insulation is tested in a test plate at a 30F delta-T with a 75F average temp for labeling purposes. Doors or SIPS or ICFs aren't easily tested that way, and they aren't constrained to those parameters in their marketing claims.

If they state an actual U-factor for that door, those claims have to meet more stringent standards than some fanciful center-paned R-value at some cherry-picked test conditions, but still doesn't have the same level of testing requirements as insulation does for R-value.

If one looks at their own test numbers, R17.66 turns out to be R7-ish in the real world. That's substantially less than the ~R9 you get for "whole wall" performance out of 2x4 wood sheathed vinyl sided studwalls filled with with low density R11 batts. See the table on this page:

http://www.haasdoor.com/about/insulation.html

Note that the test conditions was a 50F delta-T at an average temp of 43F- that's 18F outdoors, 68F indoors. Close cell polyurethane increased in performance with falling tempeartures, and unless your binned hourly average winter temp is 18F or lower (US climate zone 7 & higher) your seasonal average performance will be somewhat worse.

Note that even under those test conditioned their 700 series (1-3/4" thick) door has a calculated R of R16.18, but as an assembly tests at R6.67. (less than R7!). Their 800 series (3" thick) has a calculated R of R25.8, but a tested assembly performance of R8.34.

So, unless you live in Antarctica a 2.0" thick door is likely to do no better than R7 (a U-factor somewhere between U0.14 & U0.15 ) for a seasonal average, and that's the way you should model it from a heat load & energy use point of view.

R7 is WAY better than R1.5-2, but it's nothing like R17+.

I know the advertised R-values are terribly exagerated: that's why I said "claimed" R-17.66.  Their website also claims the doors are fully thermally broken; but I don't see how that is even structurally possible for a garage door.  They have vinyl caps at the edges of the panels to eliminate metal-to-metal contact; but that's hardly a true thermal break, IMHO.  Adding the aftermarket EPDM gaskets between the panels makes good sense to me from an air-infiltration standpoint; and I appreciate the suggestion (didn't even realize they are available).

Original quote request from the garage door company was for the 2000 series ("R-17.66").  They gave me a quote; but also suggested for my application (unheated space) to go with a 700 series door.  Price difference is $100/door ($200 total) with a difference in advertise R-value of ~1.5 (16.18 vs 17.66).  Dana, based on your post, it sounds like you're in agreement with them that it's not worth the extra money for the 2000 series vs the 700 series?  I would assume it is still, if not even more, critical to seal and insulate the rest of the structure as best I can to achieve my goal... and not just look at it as "the door R-value is really only R-6; so no use trying to go better with the walls"?  I always enjoy posting and reading here, there is such good information!