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
Musings of an Energy Nerd

Rules of Thumb for Ductless Minisplits

How to design a house that can be heated with one or two ductless minisplits — and how to operate the units once the house is occupied

Alex and Jerelyn Wilson heat their two-story Vermont home with a single ductless minisplit unit. The minisplit head (a fan-coil unit) is located on their kitchen wall.
Image Credit: Alex Wilson

Since 2008, when Carter Scott built a pioneering Massachusetts house that was heated and cooled by just two ductless minisplits, GBA has endeavored to publish reports from the field to guide people designing homes that are heated and cooled by ductless minisplits. We’ve learned a lot on this topic since 2008.

My article on Carter Scott’s approach to heating and cooling was called “Just Two Minisplits Heat and Cool the Whole House.” Since that article was published, builders, engineers, and researchers have shared their minisplit experience and data. Carter Scott has given technical presentations at several conferences (including the Westford Symposium on Building Science and NESEA’s BuildingEnergy conference); energy consultant Marc Rosenbaum has written several valuable articles on the topic for GBA (including “Minisplit Heat Pumps and Zero-Net-Energy Homes” and “Practical Design Advice for Zero-Net-Energy Homes”); and researchers Kohta Ueno and Honorata Loomis have published useful monitoring data (“Long-Term Monitoring of Mini-Split Ductless Heat Pumps in the Northeast”).

We now have enough information on the use of ductless minisplits to heat and cool cold-climate homes to set out some rules of thumb. The nine rules of thumb that I present below are based on the work of Scott, Rosenbaum, Ueno, and Loomis, to whom I am indebted.

1. Design your building to have an excellent thermal envelope

If you want to heat and cool your building with just one or two point-source heaters, you want an above-average thermal envelope. That means that the building needs a very low rate of air leakage; above-code levels of insulation; and high-performance windows.

2. Consider snow loads when placing outdoor units

If you live in snow country, your outdoor unit needs to be protected by a roof — but not a roof that inhibits air flow — or needs to be wall-mounted at least 4…

GBA Prime

This article is only available to GBA Prime Members

Sign up for a free trial and get instant access to this article as well as GBA’s complete library of premium articles and construction details.

Start Free Trial


  1. scott_tenney | | #1

    air transfer grill
    I've read about air transfer grills with a low wattage fan that kicks in when there are temperature differentials between rooms. Does anyone have experience with these? Or, is there any research re how effective these are?

    I'm wondering if this might be an alternative to supplemental electric resistance heaters in the bedrooms - that might also be effective in the cooling season.

  2. GBA Editor
    Martin Holladay | | #2

    Response to Scott Tenney
    There are two main problems with your proposal. The first is the specific heat of air, which is 0.0182 Btu/cf/°F. That means that you need a high volume of air flow to make much of a difference.

    The second problem is a low delta-T. If your bedroom is 60°F, you may want to heat it up. But the adjacent living room is probably at 70°F, so you only have a 10 F° delta-T to work with. (That's a much smaller delta-T than you get when you move air from the warm air plenum of a furnace.)

    When you do the calculations, you have to move large volumes of air to raise the temperature in the bedroom. Such a fan has its own drawbacks -- noise, for one, and the problem of drafts. It's simpler just to install a small electric-resistance heater.

  3. AntonioO | | #3

    Heat pump oversizing
    I'd heard about oversizing heat pumps before to increase efficiency. But I still don't quite understand why it works. Is it that compressors use less energy per cycle when they run at a slower speed? And if that is the case, does this apply only to a unit that is equipped with an inverter. Or should one expect a similar result for a conventional heat pump capable of running in a lower of two or three stages?

  4. exeric | | #4

    Martin, are you sure that the
    Martin, are you sure that the specific heat of air enters into the problem or at least in the way you explained it? My understanding, which is imperfect, is that a home with a given interior volume, outside air temperature, and existing level of insulation and air sealing whatever that is, will need a certain amount of BTUs. A heat pump will supply that heat, mostly in the form of warm air rather than radiant heat. If you have 2 bedrooms that aren't getting warm but your minisplit is putting out 10000 BTUS and heating the living/kitchen area to a comfortable 70 degrees then that specific heat of air is already included in that 10000 BTUs.

    Obviously to heat the bedrooms you will need more BTUs coming from the minisplit to supply them and the main living area. So if you increase the output to 15000BTUs from the single minisplit and use the in- wall fans, assuming your whole house requires 15000BTUs to heat, then I don't see why .the specific heat of air would enter into it. In other words, the specific heat of air is already included in the total of 15000 BTUs required for the whole house.

    It seems to me, and I could be wrong, that the in- wall fans would really help distribute heat. You can't ding them for poor energy efficiency when you would have to count the extra BTUs for heating the bedrooms that would always be required. It seems like its double counting.

  5. jdom | | #5

    Mini-Split Oversizing Causing Short-Cycling
    Hi Martin - Thought I'd offer up a personal experience pertaining to mini-split sizing and short-cycling. Bear with me on some background...Over the past year-and-a-half, I've been finishing a Passive House in southeast PA (certification is currently in-process). My wife and I moved in last October. The house is two floors @ 1,000 sqft/floor plus a 1,000 unheated basement. We've tested .3 ACH @ 50 Pascals, and our Passive House heat load calculation is about 8,500btu/hr. The house is heated with two single-source Mitsubishi 12,000btu mini-splits; one on the first floor and the other on the second. I had two units installed precisely for the reason stated in your post; concern about summer cooling.

    In December I installed an energy monitor on all the main circuits, and one of the first things I noticed was that the heat pumps were short-cycling most of the time; often at three to five minute intervals. This occurred whether I was running both units or only the first floor unit (I've attached a screen shot showing one example).

    Mitsubishi concluded that the problem was occurring because the heat pumps are oversized, but added even the smallest mini-splits (9,000 btus) would have resulted in the same issue (because the lower limit on both the 12,000 btu and 9,000 btu units is essentially the same). The proposed solution, which appears to have largely solved the problem, was the installation of remote thermostats on both units.

    I offer all this up to make several points. First, apparently (at least from my limited experience) oversizing may be a potential issue with mini-splits in exceptionally efficient homes. Second, the problem may not be clearly evident to occupants who aren't monitoring the actual electrical usage. And third, maybe a ducted mini-split is the better option when it comes to extremely efficient building envelopes, even though they are somewhat less efficient (as was suggested by Mitsubishi).

    I haven't been able to find any other reporting of this short-cycling issue, and didn't see anything about it in the Ueno/Loomis report. It leaves me wondering if anyone else out there has experienced the same problem.

  6. GBA Editor
    Martin Holladay | | #6

    Response to Eric Habegger (Comment #4)
    The amount of heat that a fan can move is calculated using this formula:
    Heating BTUs = (cfm of the fan) x (delta-T) x 1.08
    Where does the 1.08 come from? It's the heating BTU multiplier at sea level, determined by multiplying the number of pounds of air per cubic foot times the specific heat of air times the number of minutes in an hour (60). Another way of expressing this: the factor (1.08) is the volumetric heat capacity of air (in BTU * minute / hour * cubic feet * °F).

    We learn from this formula that the smaller the delta-T, the fewer BTUs the fan can move. The specific heat of air matters because a cubic foot of air at 70°F holds less heat than a cubic foot of water at 70°F -- which is why hydronic heating pipes have a smaller diameter than ducts.

  7. GBA Editor
    Martin Holladay | | #7

    Response to Jerome Lisuzzo (Comment #5)
    Your observation has been made by others; the solution to the problem you describe is, indeed, to install a remote thermostat.

    For more information on this issue, see Mitsubishi minisplit behaving very differently with external thermostat vs. without.

  8. GBA Editor
    Martin Holladay | | #8

    Response to Antonio Oliver (Comment #3)
    The high efficiency of ductless minisplit units under part-load conditions is a direct result of the fact that these units are inverter-driven. Single-speed air-source heat pumps without inverters don't share this feature.

    For more information on this issue, see Ductless Heat Pump Impact & Process Evaluation: Lab-Testing Report.

    Quoting from that report: "When looking at the full results from the steady-state tests, the fully variable speed nature of the [inverter-driven ductless minisplit] equipment becomes apparent. The variety of compressor and fan speeds combine to offer a huge range of input powers and output capacities to meet the space conditioning load. A clear trend from the data shows that the higher the output capacity, the lower the efficiency. Equipment performance is maximized when the loads are small. This is true for both heating and cooling. Additionally, the ability of the equipment to run at a very low speed will greatly reduce the amount of time the equipment must cycle on and off in low load situations. This on-off cycling is a performance penalty for single speed heat pump systems but, in comparison, is largely avoided with the DHP [ductless heat pump] variable speed equipment."

  9. gusfhb | | #9

    I could not see any advantage to oversizing a mini split.

    First, since they are selected for design heat load, they are already oversized over 90 percent of the time. This is added to the fact that simple heat load calculations usually over size equipment, and pretty much every assumption a HVAC contractor is going to make is going to err on the side of 'bigger is better'

    Second, in many cases in a well insulated house, they smallest available unit is already oversized

    Third, Larger units have a lower SEER rating for the same series, IOW, a 9000 btu unit may have a 27SEER, while the 12000 btu unit of the same series will have a SEER of 22. So while the unit might run at a higher efficiency percentage, it is of a lower number, thus you are not really saving actual electricity.

    Lastly, while my personal experience is mostly using them for AC, almost all of my complaints with comfort or setpoint control relate to [in my opinion] the units being oversized.

    I think the small ducted minisplits are going to excel in being sole heat source for efficient houses.

  10. user-626934 | | #10

    Load Calculations
    Martin - Perhaps, in a better world, this would go without saying....but folks ought to be reminded to get an accurate heating/cooling load calculation before considering their equipment selection. Anyone who peruses the GBA Q+A forum will notice that it's quite common to see questions about equipment sizing for a particular application in the absence of a load calc. The question from the "first responders" (typically Martin or Dana Dorsett) is always: "What are your heating/cooling loads?"

  11. user-626934 | | #11

    A caveat to #9... always, rules of thumb come with's one for #9:

    For homes with reasonably strong solar gains, and/or for any well insulated home on a mild winter day, it can be pretty easy to exceed the setpoint temperature (even with an external thermostat). In these cases, the mini-split will start to short cycle, adding small amounts of heat to the house at quite low efficiencies. For those willing to actively manage the device, there are some energy savings to be had (albeit small in most cases) by shutting it off until the sun goes down...

  12. GBA Editor
    Martin Holladay | | #12

    Response to John Semmelhack
    You wrote, "Folks ought to be reminded to get an accurate heating/cooling load calculation before considering their equipment selection."

    I agree. That's why I wrote (in Rule of Thumb #6): "Of course, it’s still important to perform a heating and cooling load calculation before specifying a minisplit."

  13. user-626934 | | #13

    ...I read too quickly, Martin.

    Carry on with the fine work. ;-)

  14. user-1017420 | | #14

    Ductless minisplit and ERV
    Hi Martin,
    No mention was made to the use of a Ductless Minisplit in conjunction with a high efficiency ERV/HRV (like a Zehnder 350).
    I presume there wouldn't be any conflict, but is there any evidence that the HRV would help evenly distribute the warm/cool air flows. You mentioned that ideally a heating unit should be used on the lower floor and a cooling unit on the upper. Wouldn't the HRV eliminate the need for 2 units?

  15. GBA Editor
    Martin Holladay | | #15

    Response to Alec Shalinsky
    As you probably know, a ductless minisplit serves a different function from an HRV or an ERV. The ductless minisplit is used for heating and cooling, while an HRV or ERV is used for ventilation.

    The air flow rates required for ventilation are low -- generally in the range of 50 cfm to 90 cfm, or perhaps 10 cfm to 20 cfm per room. These low rates of air flow aren't capable of transferring enough heat from one room to another to equalize indoor temperatures.

    For more information on this issue, see Choosing HVAC Equipment for an Energy-Efficient Home. In that article, John Straube is quoted as saying, “Ventilation air doesn’t do much to move around heat. Ten cfm of 72 degree air to a 65 degree bedroom won’t make any difference to the temperature in the bedroom at all. Open doors work better than HRV ducting.”

  16. Tom_Bassett_Dilley | | #16

    Alec, I concur with
    Alec, I concur with Martin--we thought your idea would work in a Passive House we did a few years ago (Chicago suburb, Climate Zone 5), but there just wasn't enough air flow to even out the temps. The other interesting thing we noticed in the summer is that, as the ERV runs, it dumps air just a little warmer and a little more humid into the bedrooms, so they get progressively stuffier; get a few kinds jumping around in that room, and they start complaining about the temperature. So yes, keep those bedroom doors open whenever possible, and watch the solar gain carefully for summer performance. Ultimately the owners chose to replace the Zehnder unit with a CERV to provide conditioning to the ventilation air-- it helped comfort considerably.

  17. Expert Member
    KOHTA UENO | | #17

    Oversizing of MSHPs in Heating
    Martin--great column as always; thanks for taking our research work and making sure that a wider audience gets to see it!

    Unfortunately, we never got to study short cycling effects directly in our study--due to cost constraints, we set up our kWh meters to record 5-minute total power use, and record that. Also, we didn't do anything on recording output. James Williamson and Robb Aldrich at SWA did a nice study that I *think* you summarized earlier ( As a result, we couldn't directly measure short-cycling effects.

    But I believe that Marc Rosenbaum and Robb Aldrich might have more useful information on this front.

  18. rfaesy | | #18

    Martin - In our close-to-Passivehouse 2750 sq. ft. office building in Hinesburg, VT, we are about to swap out our two Mitsubishi MXZ-3C24NAHZ2 multi-zone outdoor units that drive our five indoor heads (two MSZ-FH15NAs, one 12, and two 6s) with five matched outdoor units after experiencing significant cycling, poor turn-down rates and a seasonal COP of 1.4 last winter. Since we have a good baseline from running a "co-heating test" for 35 days over three different periods last winter with just five controlled and monitored space heaters, we know what the building heat loads and consumption are at a COP of 1.0. We should have the new outdoor units in by February 8, 2019 and will continue monitoring the building over the rest of the winter to allow us to compare performance to the multi-zone units from last winter. Keep posted on the results for a good comparison. Let me know if you'd like to follow up with a story on the results.

    1. GBA Editor
      Martin Holladay | | #19

      Interesting! GBA would love to publish a report of your findings. Email me anytime:
      martin [at] greenbuildingadvisor [dot] com.

      1. rfaesy | | #20

        I'll follow up once Bruce Harley, Andy Shapiro and I complete the study after this winter. The results will be interesting.

        1. DavidSilva | | #23

          Hi Richard, I'm about to put in a couple of multisplit systems. Your comment intrigues/concerns me that I'll be facing short cycling. How did things turn out for you?

  19. wsguerin | | #21

    Had a perhaps crazy thought about winter time heat issues: run the HRV exhaust so that it aims at the Condenser. It can then squeeze heat out of what is left of the warmer-than-outside transferred air.

    1. GBA Editor
      Martin Holladay | | #22

      Considering the volumes of air that an air-source heat pump needs to move, the tiny volume of ventilation exhaust -- usually 100 cfm or less -- which (by the way) will quickly dissipate and mix with outdoor air, won't have a measurable impact on the performance of the air-source heat pump.

  20. qofmiwok | | #24

    Can someone explain ventilation requirements? I will be placing mine at one end of a wide storage alcove next to the garage, so it will be covered and out of the snow. The installation instructions have pages of different configuration charts, but none exactly matches mine. What I see in this and other manuals though is something I don't understand. If you have 1 unit in a space like mine, it can be placed 4-10" from the corner. But once you add a second unit next to it, they both need to be 40" away from the corner. I can't figure out why this would be. If they needed 40" between them that would make more sense to me.

  21. user-1116814560 | | #25

    Great discussion re impact synergy with a ventilation system…..but with all these issues , What is the big deal about mini-splitslist, esp if ducted on new construction ?? I can get inverter technology with a conventional ducted heat pump, low temperature performance , more than adequate and precise air flow heating and cooling through properly sized trunk vents , and I can hide the vents in engineered truss flooring system , ALL at substantially lower costs and amendable to houses that are >1,000 feet per floor….I don’t get all the buzz/ appeal.

  22. jonindc | | #26

    This subject started in 2015, and here we are 6 years later. So much has changed that I don't think the advice is relevant without discussing current advances in technology, that an energy nerd would be up to date with.

    1. GBA Editor
      Martin Holladay | | #27

      On the contrary--the advice in the article is still relevant. Note, however, that the advice applies to ductless minisplits. For many homes, ducted minisplits may be a better solution.

      1. jonindc | | #28

        I have re-read the article, and I agree with you. I apologize for my comment. It was undeserved and unnecessary. Thank you for your work, and your response.

  23. SirJohnH | | #29

    Thank you for this article, Martin. It has been good food for thought. I have a question but first some context.
    I own a 95 year old, 900 sf house in Northern Colorado (zone 5b). The 425 sf basement has infloor heating - installed myself. The main floor currently has no central HVAC system. I removed it to finish the basement. Insulation seems to be decent but probably not extensive. Three roughly 100sf bedrooms and a bathroom take up one half the house. The livingroom and kitchen take up the other half. Two bedrooms off the living room. A bedroom and a bath off the kitchen. My plan has been to put in a cold climate heat pump system with mini split cassettes in each bedroom and one in the livingroom (4 total / 7k, 7k, 7k, & 18k). I feel like this might be overkill but I do like the individual room control. It seems from this article that one unit in the livingroom might be totally adequate. Since a bedroom and bathroom open into the kitchen and the kitchen is separated from the livingroom by a doorway arch, it seems like having a cassette in the kitchen would be advisable to get heat into the corner bedroom.
    Q#1: With the knowledge that has been gained since this article was written, is one of the options above a clearly better way?
    Q#2: Am I likely risking short-cycling with either of these setups?
    Q#3: Am I overlooking a better way to utilize heat pump technology to heat and cool my house? Note that I currently have no ducting in place.

    Thank you for your time!

    1. MartinHolladay | | #30

      Sir John,
      For the best comfort, I advise you to install a ducted minisplit system. Start with a careful Manual J heat load calculation; your actual design load is probably much less than you think.

      1. SirJohnH | | #31

        Martin, I ran a manual j calculation through Cool Calc. See the attachment for the results. HVAC technicians in my area are saying the recommended heating BTUs should be higher than the recommended cooling BTUs. If this is true, it creates doubt in the trustworthiness of these results. Any thoughts for me?
        An HVAC technician of mine believes that short cycling isn't a concern because modern mini splits modulate. In light of this, should I be concerned about oversizing?
        When you recommended a ducted system to me above, were you recommending ducting be run into the bedrooms? Why do you think a ducted system would be better than a ductless?
        Thank you for working through this with me!

        1. MartinHolladay | | #32

          Sir John,
          I don't have time for an in-depth answer, but here are a few quick thoughts.

          Unless the HVAC technicians in your area are intimately familiar with your house, they shouldn't make generalizations about the ratio between the design heat load and the design cooling load. Factors affecting cooling load -- for example, unshaded west-facing windows -- are very different from factors affecting heating load, so generalizations are suspect.

          It's true that inverter-based minisplits modulate. But oversizing a minisplit still results in disappointing performance. Do a search on "oversizing a minisplit" on GBA and you should get links to lots of articles and Q&A threads on the topic.

          Almost all bedrooms have such a low load that one minisplit head per bedroom results in a grossly oversized system. That's why I recommended a ducted system.

Log in or become a member to post a comment.



Recent Questions and Replies

  • |
  • |
  • |
  • |