GBA Logo horizontal Facebook LinkedIn Email Pinterest Twitter 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

Ductless minisplits

Troy Stevenson | Posted in Mechanicals on

I really like the idea of using ductless mini-splits in our new home for heating and cooling. Before I get too excited and start getting quotes, I was hoping for some opinions and feedback from those that have experience with these systems. I know it is a difficult question without knowing our heat load (Manual J in process), but any thoughts of whether it may work in our house are appreciated.

Zone 5 – Detroit, MI
2,400 SF (2 levels) + 1,200 SF basement
2×6 wall w/ cellulose + 1.5″ polyiso (r-10) on exterior
R-10 under slab
R-20 basement wall
R-75 vented roof
Very simple rectangle shape house
Pretty open first floor

We went lighter on the exterior foam than we would have liked due to buildabilty issues with unfamiliar builder/framer.

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. Troy Stevenson | | #1

    Forgot, windows are Accurate Dorwin fiberglass triple glazed. U-values ranging from .17 to .24 depending on style and orientation. Hi SHGC on south side.

  2. Kevin Dickson, MSME | | #2

    I'm just now breaking ground on a similar house, so I'll give what my decisions were, even though they won't be proven to be working in the field until the house has gone thru a year of observation.

    1. Your Manual J should show that 18,000 btuh is enough heating, which means one unit on the main floor can do the trick. There are lots of caveats, like maybe put some electric resistance baseboard heat in the upstairs bedrooms. You won't use it, because you are energy conscious, and don't mind if your bedrooms are 6 degF cooler than your living space.

    2. If you have detailed the basement insulation well (which isn't easy at grade and 8-12" up from grade) you won't need any heat in the basement. Adding a few baseboard heaters at $70 each are a cheap way to keep your builder from getting the jitters.

    3. With the minisplit on the main floor, you may find that in summer, its cooling power doesn't spread to the second floor. The cheapest solution also happens to be the most efficient solution, which is providing $99 window air conditioners where needed upstairs.

  3. Sonny Chatum | | #3

    Troy and Kevin,

    Certainly wait for Manual J, or just use BEopt for free yourself, which will give you the heating load pretty close. BEopt is powered by Energy Plus, which uses Manual J calcs in the first place. The 18000 btuh sounds a little low, to me, given that s.f. and cold winters in Detroit. However, it depends just as much or more on your infiltration, ACH.

    It sounds like ductless mini-splits could work well for you, but I doubt that you could get by with just one on the first floor--not necesaarily because of heating demand, but because of cooling demand. My initial reading and intuition, and now experience, says Kevin is correct about distribution and the 2nd floor. I did a deep energy retrofit on my house, and threw out an entire 3 ton central heat pump system in Feb of this year. Since then, I have 12kbtuh on the first floor and 9kbtuh on the second. The first floor unit can handle my heating load by itself, but the 9 unit is essential for summer.

    I disagree with having electric baseboard or window AC. Properly sized mini-split units with a floor plan open enough should be sufficient. Electric baseboard and window AC are kinda old school, at the same time you're trying to ride the cutting edge.

  4. Expert Member
    Dana Dorsett | | #4

    If BeOpt is too much data entry, run a room by room I=B=R type spreadsheet calculation to sanity check anybody's Manual-J. It'll be close enought for sizing ductless.

    The U-factor of your wall assembly is about U0.040

    The U-factor of your basement walls (only count the above grade section, down to about 18" below grade for estimation purposes) is about U0.050

    Your ceiling U-factor is about U-0.014

    Call solid wood exterior doors U0.5- insulated doors should have a published U-factor.

    Use realistic indoor & outdoor design temps. If it's Detroit, your 99% outside design temp is +5F- don't let anybody use 0F or -5F in a Manual -J "just to be sure...".

    If you assume an interior temp of 70F and a 99% design temp of 5F, that's a 65F delta-T. On a room-by-room basis set up cells in a spreadsheet sort of like:


    __________ U-factor _x_ area__x__design delta-T_=_heat loss

    Windows: ____0.17_____10.5'_______65F_________116 BTU/hr

    Doors:_______0.50_____22'_________65F________715 BTU/hr

    Wall: ______0.040_____127'_________65F________330 BTU/hr

    Ceiling:_____0.050_____154'_________65F________500 BTU/hr

    Room total:_________________________________1661 BTU/hr

    Do the same for each room- make subtotals for adjacent rooms connected with open doorways or doors that are normally left open, as potential zones for single ductless head.

    Subtract 150BTU/hr per refrigerator, 250BTU/hr per sleeping human, and add a 15-20% fudge-factor for infiltration & ventilation, but it probably won't change much.

    Any decent ductless can put out about 15,000 BTU per rated cooling ton in heating mode @ +5F. The smallest ductless heads tend to run in the 7000BTU cooling / 9000BTU-heating. If a "zone" or room doesn't have at least 5000 BTU/hr of design heat load you have options (splitting mini-duct cassette output between two adjascent or nearby rooms, resistive heat such as radiant coves or panel radiators used intermittently for temperature balance, etc.

    Even with quite a bit of glass it's unlikely that you would have more than 30K of heat load @ +5F, and either 2 3/4-1 ton mini-splits or a 2-4 head 1.5-2.5 ton multi-split can probably handle the load and distribute it reasonably, but run your spreadsheet, see how it shakes out.

  5. Troy Stevenson | | #5

    Thanks everyone. Your information is extremely helpful.

    Anyone ever encounter issues with mechanical inspectors in a ductless mini split install?

  6. Bob Coleman | | #6

    Is there a heat pump that can put out 18000 btu and still work in the 0 degree weather??
    You also need to make sure your mini split has a base pan heater, or you develop a system to take care of it - you also take a hit in performance for that sadly. Mounting above snow level and providing a little roof also help with snow.

    Two units give you more options and the small models are super efficient so there really isn't much reason to not go with both. Costs should be similar and you can just run one when needed.
    Baseboard electric makes a nice backup or covers peak issues, however a portable unit will give you flexibility if you have some circulation issues while you grow into it.

    The only inspections issues I can think of, apart from typical, might be requiring another electrical disconnect between outside unit and inside unit, an expensive 3 pole switch but simple to do, and code requires a GFCI breaker on the main unit (just like a hot tub) again a little pricey.
    You can try to go without and see if they mention anything; GFCI is easy retrofit, 3 pole not so much to make it look good. I did the GFCI breaker at the main fuse panel.

    One thing that influenced my decision, but hasn't been much of an issue as I thought, is that the Mits models don't come with a standard programmable thermostat like your typical home system. You have to manually adjust it to alter temps during the day while gone and such.
    The Fujits has a sensor on it that adjusts according to movement and it does work, but only if you pass by it often enough when you are home - sit on the couch for 2 hours and it thinks you let. You can turn it off.

  7. Kevin Dickson, MSME | | #7


    Window AC does seem old school, but is surprisingly cost-efficient:

    Removing them for the winter is a nuisance.

    A 5kbtu unit is at least 10X cheaper than an installed minisplit.

  8. Expert Member
    Dana Dorsett | | #8

    Don: "Is there a heat pump that can put out 18000 btu and still work in the 0 degree weather??"

    Sure- even ducted versions. Click on the "Heating Capacities" tab on this tool and play around with the different air handler & compressor options:

    The Mitsubishi H2i and Fujitsu XLT-H series both work fine at 0F and below.

    If you really WANT a wall thermostat with programmable setbacks etc, the Mitsubishi MHK1 works with most of their heads:

    The energy savings of setback strategies are pretty slim with mini-splits- even negative in some instances. The efficiency at part-load (idling along just maintaining room temp) is dramatically higher than at full load (going full blast while recovering from setback.) Most of the time "set and forget" is good enough, with minor tweaking with the remote.

  9. Sonny Chatum | | #9

    If you are still monitoring responses to your question, I would be interested in knowing the results of your heating load calculations.

  10. Troy Stevenson | | #10

    Actually we have engaged EnergyVanguard (Allison Bailes is principal) to do the load calcs and design the entire system. I will definitely let everyone know the results and progress as we move forward. We are very excited to have them involved.

  11. Bob Coleman | | #11

    I forgot to note the Mits has thermostat solutions, but they cost an additional half as much as there cheapest mini-split units as an add-on for the single head models.

  12. Expert Member
    Dana Dorsett | | #12

    Bob- I know- those wall-stats are pretty ridiculously priced, no doubt partly due to the very low production volumes. Most people can learn to live without 'em, eh? ;-)

  13. Troy Stevenson | | #13

    Manual J results are back. Heating load is about 37k Btuh - roughly 12k Btuh per floor (including basement). Cooling is about 15k Btuh. The cooling needs are easily met with one Mitsubishi outdoor unit, but the heating load is the issue. I am not familiar with all of the specs on the Mitsubishi equipment, but I believe that the 42k Btuh unit drops to around 30k Btuh at 17 degrees.

  14. Expert Member
    Dana Dorsett | | #14

    First, that seems like an exceptionally high Manual-J number for 3600' of conditioned space at your R-values. I would have expected well under 30K at Detroit's +5F 99% outside design temp, maybe even under 25K. Take a look at the calculations and figure out why it's that lossy (or if there are gross errors like 75F interior design temp and -10F outside design temp, etc.)

    The heating output specs on the Mitsubishi Hyper Heating units are good down to -15C/+5F. Derate to 80% at -20C/-4F, and ~70% at -25C/-13F. For the most part, ignore the cooling BTU ratings, but you generally get about 15,000 BTU heating @ +5F for every rated ton (12,000BTU) of name-plate cooling out of that series.

    eg: The popular MSZ/MUZ-FE18NA is a 1.5 tonner, good for 21,600 BTU/hr heating @ +5F, or 14,400 BTU-heating @ +5F per rated ton of cooling, and is still good for ~15,800 BTU/hr @ -13F. A pair of those would handle even the (likely overestimated) 37K load.

    The 1-ton MSZ/MUZ-FE12NA is only good for ~12.5K heating @ +5F though.

    If you went with the 2.5 ton triple you'd be good for up to 36K-out @ +5F depending on which combination of heads you used:

  15. Sonny Chatum | | #15


    Clearly you are looking for good energy performance in a cold climate, so your walls are a little low in R value, I believe, less than R30. I retrofitted my house in Maryland (warmer, about 4600 HDD) to R40 walls and I believe that's about right for here. I would design for that or more in Detroit.

    I would agree that the 37 kBtu/hr sounds a little high, but not necessarily. In addition to the walls that are a little low in R valure, the 3600 sq ft area is a lot, and only R20 in the basement, when the ground can get pretty cold in Michigan, down to 3 ft. deep, or so.

    There's been no mention of air infiltration. What infiltration rate are you using to get that heat load? Infiltration rate can make a pretty big difference in heating load. I would advise, again, to double check with BEopt software--using it is fast and easy, and particularly useful in testing various scenarios.

  16. Troy Stevenson | | #16

    Dana - Much thanks for the help on the equipment. I will have a chance to review the inputs on the load calc this week so I am hoping this number will end a little lower.

    Sonny - I agree that we went lighter on the walls than we should. However, mention exterior foam or double stud wall to anyone in this area and you just get blank stares. Gas and electricity are too cheap here. So 1.5" of polyiso on the exterior is where we compromised for 'buildability'. That said we may beef up the basement wall insulation in light of your comments. Thank you.

    We are using 1.5ACh @ 50pa as the benchmark for the load calcs. Hopefully we end lower than that.

  17. Steve Knapp CZ 3A Georgia | | #17

    After I found out that mini splits would not work with my layout, I considered ducted minis. Unfortunately, the efficiency numbers on these units was not very good. It turned out that a regular split system offered equal or better performance for about half the installed cost.

  18. Patrick Krekelberg | | #18

    Hi @Troy! What did you end up doing for your mechanical design?


Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |