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
Building Science

Oversizing a Minisplit Heat Pump System

Without knowing the proper heating and cooling loads, you can lose the benefit of a variable-capacity HVAC system

Variable-capacity minisplit heat pumps can drop to lower capacity when the load is lower. Does that make it OK to oversize them?

Recently I did a consultation with a homeowner in Michigan. He’s got a 60-year-old house with a boiler and radiant heating system that has become a burden. The pipes for the hydronic distribution system just developed their seventh leak since he’s lived in the house. To repair the leak would require tearing up the kitchen, so he’s looking for other solutions.

One possibility is going with all ductless minisplit heat pumps. One HVAC contractor proposed seven indoor units for his 2100-sq.-ft., four-bedroom house—that’s one ductless unit for each bedroom, plus three for the downstairs. The minisplits might be sized appropriately, but it would only be by accident. The contractor didn’t do or even propose doing a load calculation.

Can you oversize a minisplit?

And that gets me to the real point of this article. A lot of people seem to think that you don’t have to worry about oversizing a minisplit heat pump, whether ductless or ducted, because these systems have variable capacity. The heating or cooling output drops when the load is lower, so it’s OK to put in a unit that’s too big, they argue.

But is it true? No! Here’s why. Minisplits do ramp down in capacity as the load changes, but they don’t go all the way to zero. There’s a bottom they won’t go below.

Understanding turndown ratio

For example, the Mitsubishi FS06 wall-mounted ductless unit has a rated cooling capacity of 6000 BTU per hour and goes down to a minimum capacity of 1700 BTU per hour. That’s a turndown ratio of 28% (1700 ÷ 6000). If you put that unit in a room with a load of 1500 BTU per hour, it will bottom out nearly 100% of the time it’s running in cooling mode.

One of the great advantages of using minisplits is their variable capacity. By oversizing them, you can lose that advantage completely. That 6000 BTU per hour minisplit in a room with 1500 BTU per hour of cooling load doesn’t act at all like it has variable capacity because it’s always running at the lowest capacity.

So, yes, you absolutely can oversize a minisplit. And when you do, you end up with some of the same problems you get from oversizing conventional systems: comfort problems, poor humidity control, short cycling, and wasted money.

When a contractor proposes seven indoor ductless units for a 2100-sq.-ft. house without doing a load calculation, there’s close to a 100% chance the system will be oversized. But this problem happens with new homes that get load calculations too. You’ve got to look at the specifications for any minisplit heat pump and put in one that will take full advantage of the variable capacity.


Allison Bailes of Atlanta, Georgia, is a speaker, writer, building science consultant, and founder of Energy Vanguard. He is also the author of the Energy Vanguard Blog and is writing a book. You can follow him on Twitter at @EnergyVanguard.


  1. Randy_Williams | | #1

    Can there be an issue with cooling loads for ASHP's if the system is sized primarily for a heating load in a very cold climate? In my area, design temps are -23°f and we need at least some heat for 7 months out of the year, but cooling is still needed occasionally.

    1. charlie_sullivan | | #2

      My thinking on that is that a system with several separate compressors is better than a single compressor with multiple heads, because you can turn off all but one system and likely get enough cooling for the whole house.

    2. AndyKosick | | #6

      Turn down is absolutely critical up north. The 2 numbers I care about when selecting heat pumps are Max Heating @ 5F and Min Cooling @ 82F per the NEEP database. Max typically being higher than Rated output. In fact the database actually states the heating turndown as Max 5F/Min 47F. For FS06 it 8700 and 1600 for a ratio of 5.4, see the link. Cooling minimum are usually a little higher than heating.!/product/34394

      My goal is always to try and meet heating load while getting minimum cooling somewhere under the cooling load and this can be hard in some cases. In my opinion, even with an aggressive Manual J, using Rated output risks oversizing for overall comfort up here where heat/cooling ratios of at least 2/1. I'd be curious what Allison think of this.

      What we really need from Cold Climate manufacturers is:
      -More Turndown!!!!
      -resistance coils for slim ducted heads
      -way better controls capable of doing reheat.

      Because I'm sure there are engineering limitation to turndown I don't understand, the coils and controls an inexpensive way to allow a single system to be sized in the variable sweet spot, and then pickup those -30F days on one end and act as a dehu on the other for but a small energy penalty.

      I'll end my rant here. Any thoughts?

      1. u_value | | #12

        I am in climate zone 4A, and the approach of sizing the min-cooling HVAC output to be under the Max Cooling demand, I like. So in the case of a 28% turn-down ratio on a system with a Max Cooling demand of 1.5 tons, it puts the Max heating output at 5 tons when outdoor air temp is ~45 degrees, right around where most residential heat pumps top out. Of course, this depends on the heat pump and the performance curves. As indicated, if the heating demand is over 60 kBtu (5 tons), then back up heat will be needed.

        I guess it depends on the system being installed, but I am wondering if this kind of strategy is being implemented, whether the AHU fans will ever turn off when in cooling demand is satisfied? I know Fujitsu had a situation like this and eventually they were able to tune the system to the point that their 1 ton units can now achieve 30 SEER. Their ductless heads keep the fans running on a very low speed to be able to continue to sense air temperature etc.

        However, for many people with central systems converting to inverter-driven air-source heat pumps, will the lowest stage eventually turn off the air-handling unit once cooling demand is satisfied?

        I would note that for this "low-stage HVAC system cooling output being under max cooling demand" could theoretically produce some extremely efficient cooling output as the condenser and evaporator surface area is very large relative to the compressor power in. The cooling side would be operating essentially as a single-stage system, but it is possible to comfortably cool a home with a single stage HVAC system.

  2. jackofalltrades777 | | #3

    "Seven indoor units for his 2100-sq.-ft., four-bedroom house" is bananas. I had a Manual J done for a 1,900 sq-ft home, 3 bedroom home and they sized it for TWO ductless mini's. I also had a Manual J done for a 3,300 sq-ft, 4 bedroom home, and they sized it for FOUR ductless mini's. All of them 1-ton units. This was for a climate Zone 4B. Seven units seems like complete overkill for a home that size unless it's so poorly insulated with floor to ceiling glass. Here it's about $3k - $4k per unit, installed (Mitsubishi), so with 7 units that's a hefty $21k - $28k, depending on costs.

  3. BirchwoodBill | | #4

    I think the lesson is that you really need to do the heat balance, I.e. manual j for the building. You can oversize a boiler, mini split, furnace or AC. You also have to pay attention to the lower modulation edge of the heat source, particularly with PGH. Second is to pay attention to the latent load, I.e. humidity. Minnesota weather is tolerable if you can remove the humidity.

  4. AndyKosick | | #5

    Sorry Allison, I've got to nitpick. You stated a turn down ratio and a percentage and feel like there should be clarity here because turn down is so important especially up north.

    For the situation you stated the turn down ratio would be 3.57 (to 1 being implied). That's how the NEEP database states them and I think it's important that we're all speaking the same language. I'm going to expand on this is a response to Randy Williams above as well.

    1. maine_tyler | | #7

      Yeah I found the 28% confusing because it's saying the minimum output is 28% of the rated. So lower is better. And 100% turndown would imply 0 actual turndown. That seems backwards.

      I can just picture the confused sales pitch now: 'yeah this sweet baby has 90% turndown, while this old bird has a measley 10%.'

  5. Jon_R | | #8

    One needs to be careful to differentiate between 400% over-sizing and say 40% over-sizing. The latter will have little effect and may even be beneficial.

    I'd like to see data (vs the usual vague "it will be poor") showing which of these less-than-ideal scenarios is worse:

    a) an over-sized ductless inverter mini-split in a bedroom
    b) no heating/cooling source in a closed door bedroom

    > More Turndown!!!!
    Counter-intuitive, but this will often lead to less comfort and less efficiency. And we already have too much of "ASHP marketing specs are ruining real-world performance".

    > What we really need
    Most importantly: "good SHR (latent performance) at all operating points".
    Additionally: good low hysteresis thermostats, so temperature is tightly controlled even when equipment has to cycle.

    1. AndyKosick | | #10


      The answer to your question is:

      c) a ducted system properly balance room by room

      After initial enthusiasm, I've found ductless heads to be of limited use in residential scenarios. Allison's sunroom as perhaps the exception. A single open space with a unique and significant load profile.

      Turndown need never become a marketing spec. Just something designers use to see that a single system can do the whole job.

      I agree that latent is the crux. It would be nice if a single heat pump system could cover it. I'll emphasize controls again. The hardware on these minis is capable but the controls are over-complicated and yet also totally inadequate at the same time.

  6. Expert Member
    PETER G ENGLE PE | | #9

    >I'd like to see data showing which of these less-than-ideal scenarios is worse:
    >a) an over-sized ductless inverter mini-split in a bedroom
    >b) no heating/cooling source in a closed door bedroom

    Hard to provide data, since "worse" can have so many definitions. It seems that the most important assumption in your scenarios is how long and how frequently the bedroom door is closed. Clearly, people want privacy in bedrooms. In some houses, the doors will be closed 100% of the time. In others, not at all. With 100% closure, your worst case for option a) is probably a waste of money, up to 100% of investment in the equipment. Probably not too much of an energy penalty, possibly poorer humidity control. Worst case, option b) is poorer temperature control. If there is no significant insulation between the bedroom and the rest of the house, and lots of insulation to the exterior, comfort may not be too much of an issue. Of course, option b) also fails to meet code in most jurisdictions.

    But why are these the only options? Why did you eliminate a ducted minisplit system from consideration? These seem to be rising to the top of many designer's toolboxes for addressing closed-door bedroom conditioning at reasonable costs and with good performance overall. It also has the benefit of being closer to familiar to HVAC techs who are used to installing fully ducted systems for everything.

    1. Jon_R | | #11

      >> I'd like to see data showing which of these less-than-ideal scenarios is worse
      > Hard to provide data since "worse" can have so many definitions

      Agreed, leave the conclusion up to the user. But what we have is mostly "no data" period. Just lots of hand waving demonization of over-sizing supported by the occasional cherry picked anecdote or analogy.

      Take a common example of "over-sizing will shorten equipment life". Where's the proof that fewer compressor hours doesn't increase life? It seems to have for my 20 year old over-sized compressor. And if over-sizing does shorten life, is this a "who cares 1%" or "it will fail in a few months?". Without such data, I'd ignore the claim.

      If forced to be excessively general, right-sized ducted is a reasonable pick. But take an example: two bedrooms with lots of windows, one facing east and one west (ie, a hugely variable load ratio). No automatic zoning. With ducted, expect discomfort exceeding what would occur with two independent but over-sized units. Which leads right back to "show me the data so I can decide which is worse".

      > In others, not at all...worst case ... probably ... up to ... Probably not too much... possibly ...Worst case ...

      Exactly. Which also leads back to "provide the data so one can be sure". Not easy, but it is what is needed. Ideally, manufacturers would provide complete enough specs that it could all be modeled. But we typically can't even get "SHR at 50% load".

  7. Deleted | | #13


  8. PBP1 | | #14

    Thank you Dr. Bailes, I was visiting my friend Dr. Bales in their Usonian gem last week with glass abutting wood (in northern locale). As all mention, heat load numbers? There's nothing quite like a right-sized heat pump. In zone 5b, have 28.6 kBTU heat for 2,100 sq ft with three ducted heads. Heat load calculations (required for HERS) appear accurate and align with utility bills. I spot filled a few air leaks, made appropriate filter choices (type and size) and insulated the air handlers. Heat pump handles -15F (nightime low) to over 100 F (daytime high) just fine. Installer said they usually go with a rating slightly below the calculated heat load. And if too undersized, easy tweaks are improving ACH50, filter selection and line/handler insulation.

  9. sprockkets | | #15

    As an example, I'm somewhat happy with only needing 12k of cooling for my apartment. However, when it gets to - 20F, my 70k furnace is running continuously (most of the time it is at 50k on low stage and that's perfect for when it is only around 20 outside). If I did ductless I would need 2 36k units, minimum, just to keep warm. But at -20 they would be down around 66 percent of their capacity.
    I can't exactly trust home owners but all you can do is really ask them how well the system works that they have. Sure, do a heat calc. But with the output of mini splits and their efficiencies at negative temps, you are better off sizing for the majority of the time, and then using straight electric heat to make up the difference where needed. Heck, you should have that as a backup regardless.

  10. user-6997503 | | #16

    Hi Allison,

    We recently bought a house with a Fujitsu system: 4 ductless mini-split heads, and one outdoor unit. The house is two story, 1500 sf (850 sf on the first floor, 650 sf on the second floor). There are two bedrooms upstairs, each with a 7,000 BTU head; the first floor is all open plan, with one 15,000 BTU head and one 7,000 BTU head; the outdoor unit is 36,000 BTU rated for extra low temperature.

    I'm feeling like the units might be oversized, because it seems like we have an issue with short-cycling. My thought is that I could turn off the 7,000 BTU head on the first floor, and one of the 7,000 BTU heads on the second floor (we don't use the second bedroom much, and we leave the doors open on the second floor). Would we be doing anything detrimental to the overall system by turning off two of the heads?

    Thanks very much for any advice you can give...I want to love this system, but so far we haven't been too happy...

  11. dirkgently | | #17

    I have incredible respect for the author but I get lost in this article. The case study is to replace a heating source in Michigan, but then the discussion switches to COOLING capacity. HUH?
    I sought out this article while trying to help a friend make sense of his recent price quote for completely replacing his heating system with minisplits. I sat in on the sales visit and was not shocked when there was no manual J done. With the huge rebates being offered in MA it seemed a decent deal.
    However, with 3 tiny bedrooms each getting a 7k head, the sizing seemed way off. I did some load calcs for each room and found they needed 3,300 btu per room. Problem is there are not any heads which offer such low capacity so options are limited to the A and B scenerio noted in above comments.
    Personally I can testify that relying on an adjacent room split head does not work well (worth noting the room without head is super insulated and air sealed). I put a small electric baseboard in that room to make up the bulk of the heat and only use when needed and it heats up in minutes.

    Bedrooms DO need heat in northern climates, therefore we are stuck with installing an oversized head or possibly a low static ducted ceiling cassette which feeds 2 rooms (however this seems much more difficult in a remodel scenerio)
    8 years ago, I worked on a small Net Zero house developmet with double stud walls and the builder smartly put a split head in each bedroom knowing the risk of oversizing. He has had Zero complaints.
    I hate to say this but, it could be the sales guy I ridiculed at first got the unit sizing correct based on the available equipment, using non-obtrusive installation ideas.
    in a remodel scenerio What is a homeowner to do A or B?
    In a new construction scenerio Go with ducted air handler/heat pump

    1. user-723121 | | #18

      These are the type discussions we need to better understand the new heat pump technology and sizing calculations. That you have experience with superinsulation makes your post all the more valuable. Heat pumps when sized and installed correctly are the future of HVAC as we know it. The charging absolutely must be done to specification otherwise system failure is a great possibility.


Log in or create an account to post a comment.



Recent Questions and Replies

  • |
  • |
  • |
  • |