GBA Logo horizontal Facebook LinkedIn Email Pinterest Twitter X Instagram YouTube Icon Navigation Search Icon Main Search Icon Video Play Icon Plus Icon Minus Icon Picture icon Hamburger Icon Close Icon Sorted

Community and Q&A

Single-Zone Minisplit System

rustyshackleford | Posted in General Questions on

We’ve bought a mountain house that has only electric-resistance heat (mostly baseboard, and a couple of in-wall blower units).   Imagining electric bills will be horrifying, I’m looking to install at least one minisplit.   The house has a day basement (partially below, partially above grade), a lower floor with bedrooms, and an upper floor with living/dining room, kitchen, and office; each floor is about 850 sq-ft.

I believe if I install one minisplit on the upper floor, I will get the most bang for my buck.   Normally you want common/living space to be warmer than bedrooms; also, there are three bedrooms, and it will be rare for all 3 to be occupied (during family reunuions and such, but not for most of our use of the place).   So rely on the electric baseboard for the lower floor, and the minisplit for the upper.

My first concern is the heating design temperature.   This source:

… gives it as 15 degreesF (Watauga County NC).   But the weather station is Boone NC, which is about 1000ft lower elevation than the house.  So I think 10 degrees is the right number to use.    However, perusing a local weather site’s historical data, there seem to be an average of perhaps 10 days per winter when the low temperature is below 10 degrees.   But you’re usually not awake then, so maybe it doesn’t matter.   The average low in January appears to be about 25 degrees.

Trying to do a Manual J load calculation (, I’m not sure what to specify for the floor construction.   The tool seems geared towards an entire house, not just a room or a floor of a house.

Anyhow, it appears the answer (for heating) will be either 18Kbtu/hr or 24Kbtu/hr.   Of course, I’m tempted to go with the larger size, relying upon the ability of a minisplit to throttle down to make it work well.  I’m thinking I will probably go high-end, probably Mitsubishi.   I can do most of the install myself, and have identified a local HVAC guy who is willing to just come and do the refrigerant – he’s guessing 2hrs labor at $80/hr plus whatever extra freon he needs.

My post rambles, but I’d be grateful for folks’ thoughts.


GBA Prime

Join the leading community of building science experts

Become a GBA Prime member and get instant access to the latest developments in green building, research, and reports from the field.


  1. Expert Member
    Dana Dorsett | | #1

    A better Manual-J-ish load tool for heat pumps is the harder to screw up purpose-designed tool from BetterBuiltNW. It's free, but you do need to share an email to open an account.

    >" Of course, I’m tempted to go with the larger size, relying upon the ability of a minisplit to throttle down to make it work well."

    That is the sort of thinking that leads to sub-optimal, lower efficiency lower comfort outcomes. Modulation ranges are not infinite, and a mini-split needs at least some amount of load to run efficiently. Run the BetterBuiltNW tool, and install the SMALLEST unit that covers (or almost covers) the 99% load.

    The NEEP database is a good place to go shopping for models. They list the output & at both +5F and +17F across minimum to maximum modulation levels (as well as at +47F). If your calculated load at +10F is close to or smaller than the specified capacity at +5F, and slightly under the capacity at +17F. that's the right size. Do not bump up to the next size without comparing what the minimum modulation output & efficiency is at +47F.

    With Mitsubishi there is usually a BIG step up in cost between an 18K and 24K unit, as well as a big step in minimum modulated output @47F:!/product/34433/7/25000///0 (<< 5150 BTU/hr @ 47F)!/product/34515/7/25000///0 (<< 10,000 BTU/hr minimum @ 47F, nearly 2x)

    A unit that's running a near 100% duty cycle is mixing the air and actively heating. One that's cycling on/off at a 50% duty cycle won't be mixing nearly as much, allowing for more stratification, and delivering bigger wind chill effects.

  2. rustyshackleford | | #2

    Thanks for the reply ! That calculator looks better. And the NEEP site ... what a resource.

    In the sizing tool, I guess I should just call it an 850 sq-ft house with one room. Reduce the U-value of the floor, since it'll be above the lower (bedroom) floor, somewhat conditioned.

    When I'm looking at the NEEP tables, and seeing if my heating load lies in between the 5 degree and 17 degree values (as you suggested, but I think you had a typo), should I be looking at "maximum" or "rated" ?

  3. Expert Member
    Dana Dorsett | | #3

    >"In the sizing tool, I guess I should just call it an 850 sq-ft house with one room."

    That should work. Unless it's a VERY air-leaky 850' with lots of single-pane window you're probably within the range of most cold-climate 1-tonners, or even some of the 3/4 tonners. If it's close bumping up to a 1.25 tonner wouldn't necessarily be insane, but bumping all the way up to 1.5 or 2 tons would be on the far side of optimal.

    As much as I hate rules of thumb (that's why we have load tools, eh?) a reasonably tight 2x4 framed insulated house with clear-glass double panes usually comes in between 15-20 BTU/hr per square foot @ 0F outdoors, 70F indoors, a 70F difference. At a 60F difference (10F outside, 70F inside) the load would then be about 6/7 that much or 13-17 BTU/hr per square foot. So you're probably looking at 11-14.5K at design conditions. A typical 1-ton Mitsubishi delivers (limiting it's output by design) 13,600 BTU/hr @ 5F or higher when it's between +5F & +17F. eg:!/product/26102/7/25000///0

    That's close enough even if you're at the 14,500 BTU/hr higher end of the rule of thumb WAG. When it's colder than that you're usually in bed anyway, and during daylight hours there is enough solar gain to make up the <1000 BTU/hr difference even on a cloudy day, and as soon as a couple of humans are up on their feet and turning on some lights, that too would make up the difference. Yes, it might lose ground when it's sub-zero outside for hours during a Polar Vortex disturbance cold snap, but those events rarely persist long enough to matter.

    A typical Mitsubishi 1.25 tonner (15K) delivers about 18,000 BTU/hr at that temperature range, eg:!/product/34412/7/25000///0

    Note that with the 15K unit he minimum output @ 47F rises to 5150 BTU/hr, substantially higher than the 3700 BTU/hr of the 12K.

    Once you have developed some confidence in the design condition load calculations, set the outside design temp to +47F to estimate how much more a bigger unit would be cycling when it's in the 40s & 50s F compared to a more closely matched smaller unit.

    While 18,000 BTU/850' is a comparatively high ratio of 21 BTU/hr per foot, it's not necessarily insane oversizing, and if it's not going to be giving up much comfort or efficiency to cycling @ +47F or higher, oversizing to 18K for a 13K design load (about a 1.4x oversize factor) would give you some insurance that even during an extended deep cold snap it wouldn't lose ground. But anything beyond that would be inadvisable.

    >"Reduce the U-value of the floor, since it'll be above the lower (bedroom) floor, somewhat conditioned."

    Set the floor U factor to zero, or if the tool balks, call U0.01 or lower.

    Unless the room below is going to be more than 10F cooler than your target temp it should not count in the heat load. Unless the floor itself is heated, the temp at the floor is almost always going to be 3-5F cooler than the average room temp, and the ceiling is going to be 3-5F warmer, all due to normal stratification. So the ceiling of a fairly cool 60F room is going to be 63F-65F, and the floor of the 70F room above is going to be 65F-67F. Allowing even R2 (U0.50) using the horizontal air films plus the material of the flooring/subfloor, the heat transfer at such a low temperature difference is "in the statistical noise" relative to inherent errors in the U-factor estimates for walls/ceilings/floors that have a significant temperature difference.

    The U-factor of the floor only matters significantly when it's over completely uninsulated ventilated crawlspace or open pier foundation.

    >"...should I be looking at "maximum" or "rated" ?"

    The "rated" value is not particularly relevant- capacity is all about the maximum output.

    While there can be modest derating from the maximum to account for defrost cycles, any derating would normally be pretty small when the design temp is <+15F. Only when weather patterns at that location has a high prevalence of rime-icing at or near the design temperature would it be prudent to derate the capacity for defrost cycles.

    The "rated" numbers are only the modulated output level used in the AHRI testing for efficiency. The AHRI test isn't super-relevant for sizing, since there are other factors that make it different from normal operation, such as requiring the inverter to only operate at 60Hz for the test, which is a wholly artificial constraint. In normal operation inverter-drive compressors vary in frequency over a wide range (usually more than 3:1) to better optimize efficiency & capacity. The NEEP data for min/max explicitly removes the fixed frequency constraint, and it also ignores defrost derating. Going with the maximum number is just fine for sizing the equipment, with rare exceptions. Using the AHRI "rated" number is almost never the right thing to do, since in most cases it will sub-optimally oversize for the average load condition.

    BTW: Adjusting the design temperature for altitude by -5F per 1000' of elevation is excessive. From a purely adiabatic point of view -3F/1000' would be more accurate. Local conditions can skew that a lot though, say when you're 1000' with a clear sky above an often fog-bound valley, etc. Only rarely would it be big enough to affect the equipment sizing.

  4. rustyshackleford | | #4

    Thanks again ! Lots to disgest here ...

    I'm coming up with 18,400 btu/hr (10 degree design). Huge amount of glass on the west wall, and cathedral ceilings. I'm not there now, so only guessing on dimensions. I don't really know for sure how good the insulation is either. Was built in 2001, so probably not too bad though.

    That's dividing the floor into two rooms though (see floor plan here). Maybe I can rely on the baseboard for the bedroom. The electrical panel is FULL though, so I'm going to have to disconnect a baseboard circuit to open up one for the minisplit. So need to see if bedroom is on a different circuit than the electric-resistance units in the big room; if not, gonna have to rely on the minisplit to heat the smaller room too.

  5. rustyshackleford | | #5

    I'm wondering if a floor-mount is better for me, since I'll be doing very little cooling (house is at 4200ft, and most people there don't even have A/C at all). For some reason the floor-mount systems seem to be a LOT more expensive though. I assume I want to mount indoor air handler on that heavily-windowed west wall too (same idea as conventional HVAC registers along outside walls) and I probably can't fit a floor-mounted unit there.

Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |