R10 under a radiant slab with 47F subsoil is enough from a "will it work reasonably" point of view, and at standard code-minimum construction would put the seasoanal heat loss per square foot to the great-earthen-heat-sink at a comparable level as the rest of the house's loss to the great outdoors.

But depending on your fuel type & system efficiency there's a net-present-value financial argument for more. (Especially if using Type-I or Type-II EPS instead of 1.5lb XPS.) In an ~ R26-28 ICF house it will take more than R15 to make the heat loss out of the slab into 47F subsoil comparable per square foot seasonal clear-wall losses out of the above grade walls. The PassiveHouse models are perhaps a bit too messy regarding soil types, etc, but it's been instructive to me that the trend has been to make the sub-slab R nearly identical to the clear-wall R for PassiveHouse designs implemented in US climate zones 5 & 6. Even in the Lafayette Louisiana PassiveHouse they had R17 under the slab(!), and R21 on the crawlspace/basement walls.

True, joysen's project is insulated to a bit less than half what it would take to do a PassiveHouse, but being consistent and going a bit less than half for under-slab what a PassiveHouse would call for in MN would be in the R25-R30 range.

yes, but passivehouse specs are only economical if you are actually going to do a passivehouse. there is no economic argument for them if you actually are going to have a heating system as well.

Which is why I'm not recommending anything LIKE PassiveHouse levels of sub-slab insulation for an R27-ish house, but rather 1/3-1/2 of PassiveHouse levels to bring the heat loss out of the slab in line with the losses from the rest of the house.

If it's not cost-effective under the slab, it's not cost effective in the ICF walls either. 6" of Type-I EPS costs more than 2" of XPS, but in a 15-25 year financial analysis the cost delta is recoverable in fuel savings. If the heating fuel is propane or oil it may turn NPV+ even sooner.

My mistake, I misunderstood where you were going.

I do have to note we did "code standard" R10 under our shop slab. the lower level, which is more than 50% walkout and with R20 in the foundation wall and about R24 for all exposed walls with R20 edge insulation, stayed very cool all summer, despite the fact that it also has two big leaky garage doors and we have an ERV system redistributing some heat from upstairs to the lower level I can only attribute this to two factors.

1. some heat will migrate up a stairway to the upper level. but the doors were often shut. I think this was a small consideration.

2. I suspect we lost more through the slab then many would expect, including the basic "siggy model" I am using for my slab loss estimation.

I *really* don't want to have to paw through the passivehouse calculations to reverse engineer a decent slab loss estimation though. I don't believe even german passivhaus engineers really know the depth, direction and speed of their water tables or the exact conductivity component of the soil on their jobsites. wild stuff in there. Have you found any good basic trends in their calculator?

I can't claim to have dug into it deeply. I too am loathe to bury myself in that much calculatin' just to come up with an implied precision that may or may not be accurate to more than 1.5 significant digits without doing all of the field engineering to verify the conditions. I'm not sure that ANYONE actually takes it that far for a 1-off, house, but if one were doing a community development of 100 PassiveHouse structures on a fairly flat & consistent parcel it might be worth digging in a bit further. For a single PassiveHouse, making the slab insulation something similar to that of the walls is almost certainly going to be enough, and would cost less than the engineering study to that might tell you you only needed 3/4 the R value that you used.

For my purposes, for non-PassiveHouse levels of insulation (but better than code) it seems reasonable to make the simplifying assumption that over R10 it will be similar to insulating against a massive heat sink at the deep subsoil temp, of sufficient thermal mass that it won't change a whole lot over several years. It will over-estimate the loss every time, but for houses of reasonable footprint it would not be a 2x overshoot, or even a 1.5x. On some big square slab 100' x 100' or larger the center-slab error may be large enough to matter though.