it's O2 barrier PEX throughout, stop wasting your time on using a high heat water tube heat exchanger tankless as a boiler- use a fire-tube condensing boiler instead, which 95 times out of 20 can be pumped direct, due to the very low pumping head of the heat exchanger.

An all stainless Westinghouse WBRUNG-080W / HTP UFT-080W (same boiler different sales channel- the tech support line for Westinghouse even rings in HTP's office). At internet pricing they run about $1500 - $1700. There is a lot more design forgiveness than with any tankless, and it's designed to take it, probably outlasting a Takagi by more than 2x. That boiler modulates down to 7600 BTU/hr out in condensing mode, and can handle a very wide range of flows.

You still have to get a handle on the heat load, and you've offered next to zero detail. You don't say what R-value the ICF is, or how much of what type of insulation is on each side. You also don't say how many windows of what U-factor/type/size, how much insulation there is under the slab, how much fluff there is in the attic, etc, or even what your

__99% outside design temp__ is, which makes it difficult to even put an upper bound on it. But let's give it a shot, using an I=B=R type calculation, which for an ICF house is going to overestimate by more than 15%, possibly more than 25%.

Assumptions:

1> The window/floor area is about 15%, and the windows are all U0.32 (single l0w-E argon)

2> Each house has two solid wood exterior doors (about U0.5), four door total.

3> Exterior wall height is 9'

4> The attic has R49 (current code min) or about U0.026.

5> The ICF is 2.25" + 2.25" of Type 2 EPS, R19 total for just the but with interor & exterior finish coatings comes in at U0.045 (steady state performance with no credit for mass effects.

6>The house is outside of Traverse City (99% design temp +2F) with a local design temp of 0F, indoor design temp is 68F, for a delta-T of 68F.

7> The houses are extremely air tight, and use heat recovery ventilation.

The basic calculation is:

U-factor x Area x Delta-T= BTU/hr

With the 50' diameter house there is 1963' of floor area and U0.026 ceiling, for ceiling losses of:

.....U0.026 x 1963' x 68F= 3471 BTU/hr

A 15% window/floor fraction makes for 0.15 x 1963= 294 square feet of U0.32 window for window losses of:

.....U0.32 x 294' x 68F= 6397 BTU/hr

With a pair of exterior 22 square foot U0.5 doors that's 44 square feet, and door losses of:

.....U0.5 x 44' x 68F= 1496 BTU/hr

The perimeter is 157', and 9' tall for 1413' of gross wall area, less 44' of door and 294' of window leaves 1075' of U0.045 wall for losses of:

..... U0.045 x 1075' x 68F= 3290 BTU/hr.

Add it all up and it's 14,654 BTU/hr, which is a whole lot higher than even the smaller Takagi's minimum firing rate, and it's an overstatement of reality in spite of not calculating in slab losses of the radiant through whatever foam you have down there.

Even at a design load of 14,654/1963= 7BTU/hr per square foot of radiant floor, even under the 0F outdoors design condition the floor will only be about 3F warmer than the average room temp. When it's above freezing out it'll be only about 1.5F warmer than room temp- barely noticeable if the slab is insulated (and it had better be!)

The losses of the smaller house will be of course smaller, but not exactly proportional due to having a higher surface area to floor area ratio, and the unknown construction of the 150' addition. But for yuks let's say it's proportional. With a 30' diameter the floor area of the house-proper is 707', plus 150' of addition makes 857'. If proportional the load of that zone is on the order of 14,654 x (857/1963)= 6385 BTU/hr.

So total load you're looking at about 21,000 BTU/hr, but the smaller zone's DESIGN load is barely more than half the minimum output of the smaller Takagi, and only 85% of the min-fire output of a UFT/WBRUNG-080W boiler. With the Takagi that means it's going to cycle on zone calls no matter what. But if under outdoor reset control (not an option with a tankless) the boiler would run very long, nearly continuous highly efficient burns even when it's only in the 40s F outside, even if the IBR calc is overshooting reality by 25% (which it might be.)

Go ahead and run your own IBR load numbers for each house/zone separately, using the actual window area & U-factors, and attic R etc. If you find a competent engineer to run the real load numbers we'll see just how much overshoot there is relative to your IBR calculation (not my WAG based on a bunch of assumptions.)

Either Grundfos Alphas or Taco Viridians would be appropriate for the application, but they aren't infinitely tune-able- you still have to pick the right ones. It's good to calculate the pumping head on the radiation and (if water -tube tankless) figure out what it's going to need for a primary pump. At 2gpm and a 20F temperature rise it already has the combined loads of both houses covered, so something between 1-2 gpm and would do it without over-pumping the hell out of the tankless.