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

Double-Sizing a Minisplit

samvi | Posted in General Questions on

Has anyone considered double sizing a minisplit rather than choosing hyperheat? Like many, ive compared performance ratings of minisplit heat pumps. Hyperheat is almost a no-brainer in northern states or so i thought. If you compare the COP of a top performing 9k evi minisplit with an average 18k unit… see graph below. The COP of the average 18k unit blows the doors off a top of the line unit in heating down to about 10°F. admittedly there are three things to think about in doing this. 1)temps below 10°F …most of us use backup below 5°F anyway. 2) oversizing in summer, some installs can turn one unit off to balance load and capacity. 3) Price. Is an average 18k cheaper than 9khyper? In most cases, yes. You are not only getting far more btus with the 18k, you are getting cheaper btus since the COP is higher. The 18k has a larger outdoor radiator to collect heat so it makes sense how it could perform better at certain temps. The high performance unit in the graph is a gree saphire and the 18k is just a typical 18k i found on neep. Let me know your 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. yesimon | | #1

    You have 2 problems here:

    1. Using COP as your primary metric. COP is easy to understand so people love it, but you have to understand it is measured based on "rated output" which is usually close to maximum output for a unit. Hence rated COP at 47F has very little value because you are very unlikely running at maximum output when it is 47F outside. On the flip side, the marginal COP advantage for a hyperheat is magnified at colder outdoor temps because you have much more heat loss and consuming much more energy when it is colder out. Assuming the rated COP for "normal" at 17F is better than non-hyperheat is again incorrect because the normal will have dropoff while the hyperheat will be modulating at a lower level than it is capable of, thus usually having a better COP than rated. If only someone could create a metric that integrates COP over an entire season's worth of heating demand with different outdoor temperatures!

    2. Comparing different size units/oversizing units. Oversizing is really bad for inverter-based heat pumps because all of the efficiency gain comes from being able to modulate low and sip power, instead of cycling on-off. A multi-split might only be able to modulate down to 50%, while a single-head unit might go down to 10-20%. The more your oversize, the bigger the minimum output and the more you lose the benefits and efficiency of variable speed. To extend this thought experiment: you can get a 60k btu "normal" mini split with the same COP numbers as the 18k. That's not better!

    To dig really deep, you are correct that hyperheat incurs an efficiency penalty compared to non-hyperheat at moderate temperatures (40-50F), but the bulk of the heating load and energy consumption does not come from moderate temperatures so the trade-off is worth it in colder climates. If you live in climate Region III or below then you probably shouldn't get hyperheat.

    1. samvi | | #2

      I agree, its not the best choice for every application. We have two 9k hyperheats and love em. Most of the work they do is in the 20° to 40°F range (why this looked good to me).
      Thanks for the input!

  2. AndyCD | | #3

    A good question, thinking outside the right-sized box! Your assertion that "most of us use backup heat below 5 degrees" is wrong, and offers no support for your argument. For a decade I've had no backup heat for my two 9K units in CZ5. I suspect SOME of us use backup heat, but as new homes are built to modern codes--and as minisplits continue to prove themselves--the number is dwindling. Broadly speaking, I'll also just point you to every Q&A thread on GBA about minisplit comfort when oversizing is suspected or confirmed. In the real world, never mind COP--comfort is what we're wanting and oversizing is really bad at that.

  3. gusfhb | | #4

    Mini splits are a manufactured item, built for a purpose. As such the manufacturer makes assumptions about their use and installation
    A larger unit has a larger fan which will not slow down enough for comfort in a too small room. It will affect the units ability to keep the room temperature stable.
    We already have to size units well oversize for design temps, meaning in more common temps they are larger than need be.

    While smaller units are more expensive per output BTU, I see no evidence that a cheap 18k mitsubishi unit is cheaper than the most expensive 9k mitsubishi . The most expensive mitsubishi 9k is about 1800 bucks right now, and any unit I can find at twice that size is a much higher price.

  4. walta100 | | #5

    Part of the answer is about you location what % of the time is your weather below 10°F? for me it is a very small number maybe 5%

    For me oversizing for cooling load would be a no go.


  5. Expert Member
    DCcontrarian | | #6

    This is actually an interesting question. With a traditional heating system, a BTU is a BTU, efficiency doesn't depend upon conditions. So you can just use very gross calculations like degree-days and be done.

    With a heat pump, efficiency goes down as the outside temperature goes down, so measures like degree-days are less useful, you really need to know how many hours a year to expect at each temperature to model energy usage. Heat pump capacity also decreases with the temperature, and if you get to the point where the heating load exceeds the capacity you have to supplement with resistive heat (presumably), with a COP of 1.0.

    At the same time most heat pumps have a higher COP running at part capacity than at full capacity. So when you do the degree-by-degree modeling you can't just use the COP at rated capacity at that temperature, you have to adjust the COP for the amount of utilization of capacity at that temperature.

    What you would end up with is a spreadsheet where you list every degree in your heating range, how many hours a year you would expect at that temperature, what you'd expect the heating load to be, what you'd expect the COP to be, and what you'd expect the heat pump capacity to be and whether you'd have to use supplemental heat, and how much. You could then model your energy usage for the year.

    You could them substitute in different heat pumps and model the energy usage for each. With the price of electricity -- including assumptions about future costs -- and the cost of capital, you could see which equipment has the lowest lifetime total cost of ownership.

    Out of intellectual curiosity, I actually did part of this analysis in response to a question here about the viability of a heat pump in Whitehorse. I didn't get around to adjusting COP for capacity. You can see it here:

    That covers heating. Then you'd have to look at the cost of cooling, and whether the selected equipment would give satisfactory comfort during cooling season ...

    1. samvi | | #7

      DC, yeah you arent missing much detail, thanks for the link! I must not understand your "break-even-temp" i just did the math on a friend's new install today. A c&h 18kHH, minimum cop 2.2 , elec 9.9¢/kw propane at $1.83, gas furnace 80% afue. 1million btu=$12.79 vs lpg at $25.69 for 1million btu. They will get cold when heat load exceeds capacity before they ever hit break even temp. All the comments are awesome and contribute to a bigger picture! Much thanks. I agree theres no value in the concept with the 18k mits vs 9k h2i.

      Whats a bit difficult for me at times is the machines that arent so clear about specs. For example, we have a 12k pioneer heating a 3 car garage to 61F. It was 18°F oat producing 162°discharge temp, iat was very close to setpoint, fan,auto. This was sold as a heat down to -13F unit. HVAC rule-of-thumb, disch temp 410a is 100-110 above ambient. That would be 130F tops if it were a typical compressor. So they did something to it that makes it perform better.

      As several comments in this thread point out, the COP varies under conditions and theres the defrost downtime that i dont think are figured in to the neep numbers. Its a very complex analysis between one versus another. It would be a lot easier to compare machines if they just came out and described whats under the hood. Evi or not. Radiator square inch, air flow, pump output etc. Instead of all these BS metrics that can be wrangled by the well versed. From my view, most brochure do not intentionally mislead anyone but i do see important information omitted at times. Again, awesome feedback!

      1. Expert Member
        DCcontrarian | | #8

        "Break even temperature" is the temperature at which occupant activity -- cooking, appliance use, and human metabolism -- equals the heating load, and above which no heating is needed. It's a crude assumption.

        1. samvi | | #9

          Oh That makes sense HA! Thanks for spelling that out.

Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |