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

Load calculation — ductless minisplit

jserr68594 | Posted in General Questions on


I have attached a drawing of my apartment that I am sizing a mini split for. I am thinking of using an 18,000 BTU two zone outdoor unit with either a 9K or 12K head unit in the kitchen plus a 9K unit in the bedroom area. I have had varying opinions on what size outdoor unit to use as well as which head units to go with. I was hoping to have someone with HVAC experience do a load calculation.

I am near Toledo, Ohio. The apartment is over a crawl space. I have R11-R13 wall insulation and 10″ of cellulose blown in to the attic. Both windows and patio door are brand new.

Thanks in advance!

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. iLikeDirt | | #1

    What was the question? Regardless, the answer is probably something along the lines of, "You need to do a load calculation before you can select the equipment." :-)

    That said, 18,000 BTUs for 455 square feet is ridiculous, probably at least 2x too big given the construction and your climate.

  2. jserr68594 | | #2

    I am looking for a load calculation, as stated in my 1st post. The smallest 2 zone minisplit manufactured by Midea (The company I am planning on using) is 18,000 BTU. It does however use inverter technology and allow for various BTU rates depending on need.

    NOMINAL COOLING (BTU/H) 18000 (3995-19489)
    NOMINAL HEATING (BTU/H) 18000 (2539-24249)

  3. GBA Editor
    Martin Holladay | | #3

    Here is the information that Justin forgot to include: He meant to tell us that he has already posted a thread about this house. Here is the link to where this whole conversation started: Minisplit Questions (Wall Mount Vs. Ducted, Zoning).

    [Note to Justin: Once you start a Q&A thread, it's best to post all subsequent questions on the same thread, rather than starting a new thread.]

  4. GBA Editor
    Martin Holladay | | #4

    On the other Q&A page where this thread started, Dana Dorsett kindly gave you some information on your likely load.

    That said, if you want a real load calculation, you have to do the math. Either (a) hire an energy rater to perform the Manual J, or (b) do the work yourself.

    If you want to avoid the cost of the Manual J software, but you are still a DIY kind of guy, you can use the old I=B=R pencil and paper method. Here are some articles on the topic:

    How to Perform a Heat-Loss Calculation — Part 1

    How to Perform a Heat-Loss Calculation — Part 2

    Calculating Cooling Loads

  5. Dana1 | | #5

    I sketched out a fairly complete load calculation on your last thread. To get any clost you need to be VERY specific about the square footage of wall area, and the published U-factors of the windows & doors. The true heat load is going to come in around 5000 BTU/hr @ 3F outdoors, give or take 1000 BTU/hr.

    When the minimum speed is higher than the immediate load the thing cycles on/off which makes it less efficient (and less comfortable.) The minimum speed on the 12K head is higher than the minimum speed on the 9k head, and even with just two 9k heads it is more than 2x oversized for the load. If you're going with a Midea multi-split, a pair of 9K heads is better than a 9K + 12K.

    The HSPF test is based on a crude linear approximation of miinium & maximum capacity at +47F, & +62F,, and the capacity is again tested at +17F. But the ACTUAL capacity a minimum speed is not linear- it's fairly flat to slighlyt alling between 35-45F, then rises a bit as temps fall from ~35F to 25F before falling again. Unless your load at 35F is above the minimum output, it may be cycling even when it's in the low 20sF, and will not make it's HSPF numbers due to the cycling losses. This characteristic is a pretty severe hit for heat pumps with low turn-down ratios but not so much with the Midea compressor, which has a turn down ratio of nearly 19.5:1, as long as it's sized no more than 1.5x for the load. But when it's oversized by nearly 3x ((the thing puts out about 14,000 bTU/hr @ +5F- see: ) it matters. If your load is 5K @ 3F your load is 2500 BTU/hr @ 38F, which is near the low point of the minimum capacity curve. If your load isn't bigger than that at 38F, depending on how much the minimum capacity rises below that it's in danger of cycling even into the 20sF. Putting the smallest and lowest-modulating heads possible gives it the best shot of at least coming close to making it's HSPF numbers.

    For a more detailed discussion of how the HSPF testing is affected by this non-linearity, exaggerating the as-used HSPF of inverter drive heat pumps (particularly those with low turn down ratios or high minimum speeds) see this:

  6. jserr68594 | | #6


    You did provide a lot of help and info for me. I really appreciate it and can tell that you know your stuff! What has been frustrating for me is I have had 3 contractors come out and give me quotes (None have done a load calculation that I have been aware of). Two of them wanted to use a 24K-27K outdoor unit with a 12K and 9K indoor. One wanted to use a 18K w/ two 7K head units. I was getting different answers from everyone. After seeing your post, I was pretty confident that smaller was better.

    I did post my questions on another forum as well. Someone with load calculating software ran my figures. I attached those reports to this post. Thanks again for everyone's help!

  7. Expert Member
    Dana Dorsett | | #7

    I didn't see mention any 7K heads on the Midea short sheet, but if they have them that would be preferable to even the 9K heads at your load.

    Picking apart the attaged Manual-J:

    The 1700BTU/hr of floor loss adder is just ridiculous, implying a pier foundation with huge air movement, and a crawlspace that is actually drops to 3F when it's 3F outside, which isn't very likely at all.

    The 740 square feet of exterior wall with the 4462 BTU/hr of loss means they presumed the east side and garage walls are all truly exterior and at 3F under design conditions, which is simply not the case.

    They subtracted ZERO BTU/hr for plug loads & occupants? (They added some for the cooling load, which was right.)

    Unless the place has an atmospheric drafted gas fired water heater, the 20cfm (1433 BTU/hr ) of combustion air for the gas water heater isn't legit.

    The 2805 BTU/hr of infiltration (or even my prior 2106 WAG) is on the high side for a brick-clad wall & brand new windows

    Code min interior design temp is 68F, they used 70F, which by itself adds about 3% to the numbers.

    To do a Manual-J right it has to be AGGRESSIVE, taking EVERY opportunity to come up with a lower number when it can be rationalized by the facts, and even then it usually ends up higher than measured reality.

    From ACCA Manual-J version #8:

    Manual J calculations should be aggressive, which means that the designer should take full advantage of legitimate opportunities to minimize the size of estimated loads. In this regard, the practice of manipulating the outdoor design temperature, not taking full credit for efficient construction features, ignoring internal and external window shading devices and then applying an arbitrary "safety factor" is indefensible.”

    No additional safety factors are required when load estimates are based on accurate information pertaining to the envelope construction and duct system efficiency. Large errors are possible if there is uncertainty about insulation levels, fenestration performance, envelope tightness or the efficiency of the duct runs installed in the unconditioned space.

    And, even with rapidly accumulating the errors to the high side piling on in the none-too-aggressive Manual-J inflating the load to an insanely high 13,385 BTU/hr the thing STILL comes in under the 13,800 output of the 18K Midea with a pair of 9K heads at 70F room temp (an HSPF test condition) and +5F outdoor temp, which is the same temperature difference as a 68F code-min interior temp and a +3F 99% outside design temp.

    Bottom line, it really is MORE than enough mini-split to heat the place, with a SUBSTANTIAL amount of margin, given the magnitude of high-side errors in the Manual-J.

  8. Expert Member
    Dana Dorsett | | #8

    BTW:If the 7K heads were a Fujitsu proposal, note that the AOU18RLXFZ & AOU18RLXFZH units don't modulate lower than 6800 BTU/hr, (more than 2x that of the Midea!), which is probably more than your design heat load, and will result in much higher cycling losses. That means the Fujitsu multi-splits would come nowhere NEAR hitting their HSPF numbers at your house! Look up the MINIMUM modulated numbers on any proposed solution- if it's not under 3000BTU/hr it's not a very good fit for your loads, and that's a tough bill to fill with 2-head units.

    A pair of Mitsubishi -FH06NA half-tonners (two outdoor units, one per head) rather than a multi-split would probably do better for your situation. They modulate down to 1600 BTU/hr @ 47F each, but can deliver 8700BTU/hr into a 70F room at +5F. While you're away you can turn the one in the bedroom completely off, turn the other one down to it's minimum 60F and it would heat the place just fine without much cycling at all, and would probably hit pretty close to it's HSPF 13+ rating:

    If you're there during the winter and the bedroom is too cold, you can turn the bedroom unit on. If you're the type who likes to sleep in a cooler room you might just install one FH06NA (or an FH09NA) in the main area, and use a space heater to deal with design day when you're there. (That would not fully meet code unless the space heater for the bedroom was hard wired, operating on a thermostat.) In a 60F room it would have well over 10,000 BTU/hr of capacity @ +5F outdoors, but if that's too risky, for about $200 more you can get the FH09, which is good for 10,900 BTU/hr @ +5F in a 70F room:

  9. jserr68594 | | #9

    Dana, it looks like the 18,000 btu with two 9K head units is the way that I'm going to go. The 7000 BTU units that I referenced above were of a different brand and we're much more expensive than Midea. I want to stick with Midea for that reason.

    What do you think of console head units versus the high wall units? The console design may work out better for my floor plan. I would put the units near the floor. Here is the model I am looking at:


  10. Expert Member
    Dana Dorsett | | #10

    The floor mount consoles are preferred for heating dominated climates, since they draw in the coolest air in the room near the floor rather than warmer ceiling air. A few degrees cooler air makes it bit more efficient, and gives it a bit more capacity too.

  11. jserr68594 | | #11


    Please take a look at the revised drawing. I added in the locations for the indoor and outdoor units. I depicted where I could put the potential console and wall evaporators. What do you think of where I located them? Which do you think will serve the kitchen/living area better (the console or the high wall)? Thank you!

  12. Expert Member
    Dana Dorsett | | #12

    Per the short-form specs in the Midea brochure, the combined maximum refrigerant line length is 98 feet, no more than 66 feet for the longest run, and no shorter than 10 feet on any given run, and a maximum vertical difference of 33 feet between the compressor and either indoor unit. That probably isn't going to be too tough to meet.

    Be sure to mount the thing on either wall brackets or a stand of some sort well above the historical snow drift dept in your area (not that you EVER get lake effect snows of any consequence in Toledo ! :-) ) , and protected by roof overhangs (even if you have to build on a shed roof above it, providing adequate clearance. I couldn't find an installation manual online with all of the minimum clearance information, but don't park the outdoor unit backed up too close to the wall- give it enough room to get adequate air flow, and to be able to inspect & vacuum out the coils.

    The high mount indoor unit needs to not interfere with the door, and should have at least a foot of clearance from the ceiling. Some vendors' units specify that it can be as close as 4" to the ceiling, but doubling whatever the spec happens to be will usually work better than taking it right up to the minimum. Don't tuck a high-mount unit tight to an inside corner either (as drawn in option B.) The option-B high wall unit would probably work better if you can install it blowing toward the hall, and mounted at least 12-18" away from the corner.

  13. jserr68594 | | #13


    I could install the high wall unit in the living area above the 2x4 window. I could set it about 6" away from the ceiling there due to the window. The apartment has 8' ceilings. That would put it blowing towards the hall. Would it be better to do this or mount the unit on the garage wall a foot from the ceiling?

    I also wanted to run this particular scenario by you. I was wondering your thoughts of using a 12K unit in the living room to heat/cool the whole unit? I would still install the 9K in the bedroom, but it would rarely get used. This is a vacation place that I would heat when I am not there, but I figure the one 12K may be more efficient running on its own. I would only turn on the 9K when I am sleeping and have the door shut (Which is on the weekends in the summer). Thanks!

  14. Dana1 | | #14

    The capacity of a single 9K head (10,000 BTU/hr @ +5F) is enough to heat the whole place, no need to bump it to the 12K.

    If the installation manual specifies that 6" clearance to the ceiling is OK (probably does), above the north window blowing south would be preferable than mounting it on the garage wall blowing east.

    The down side to mounting it that high is a greater offset between the average room temperature and the temp that the head is sensing from it's incoming air, but if you're not there even a 10F offset (unlikely) wouldn't be of any concern. When occupied you just bump it to where you're comfortable, knowing that the remote's setpoint number varies from the average room temp.

  15. jserr68594 | | #15

    I've looked at the manual. It doesn't state clearances for the head unit surprisingly. It gives very specific measurements for the outdoor unit but nothing for the indoor.

    I spoke to two different people at Midea. One told me 6" and the other told me 12" for the minimum. I can only assume since they do not publish in their manual, that it must not be critical.

    Regarding the 9K VS 12K, do you think the 9K unit will cool the whole place on a hot summer day with the stove running and doors getting opened occasionally?

  16. Expert Member
    Dana Dorsett | | #16

    A 9K mini-split on a north facing room (limited solar gain) is plenty. Your friend who ran the Manual-J numbers came up with only 8485 BTU/hr sensible gain, 1511 BTU/hr latent for ~10K total for the whole condo. And that included some gains that you don't actually have, since some of the west wall is shaded gy the garage, the infiltration rates are likely lower, you don't have a gas-fired water heater, etc.

    You're probably looking at less than 8000 BTU/hr of peak load for the entire condo, and that will be mostly in the bedroom, since the bedroom has both south and west facing windows soaking up late-in-the-day solar gains. According to the Manual-J those south and west windows combined contribute about 1800 BTU/hr to the peak load (credible), and most of your wall gains in the bedroom are probably another 400-500 BTU/hr (the 800 BTU/ht total is higher than reality, but the wall gains are primarily the unshaded south side.)

    Clearly the instantaneous load of running 4 burners on the stove while baking cookies on a 90F (the 1% outside design temp) day in July would overwhelm the capacity of the 9K, head, but you won't be measurably better off with a 12K head. A 9K head would be more appropriate for your average cooling load, more than covering any north-side gains plus the 2400 BTU/hr of plug loads and ~300 BTU/hr of combined ceiling & floor (exaggerated floor load) they included in the Manual J.

    Even with the 2400 BTU/hr of appliances...

    ...4 people hanging out adding up to maybe-1200 BTU/hr of load... exaggerated ceiling + wall + floor load of 1500 BTU/hr... (more than included in the Manual-J) the 1600 BTU/hr of north facing window (per the Manual-J)... still only comes to 6700 BTU/hr, which is nowhere near the 11,000 BTU/hr maximum cooling capacity of their 9K head, or even the nominal 9000 BTU/hr at which it's SEER was tested.

    Stop trying to oversize it by more than it already is, OK? :-)

    These things are designed to run super-efficiently with a "set and forget" strategy, letting it modulate with load. The bigger the oversizing factor, the less modulation range you'll have, and it NEEDS that modulation range to run super-efficiently. If you oversize it too much you will have re-invented the non-modulating on/off heat pump, which is fine, but it's less efficient and less comfortable.

  17. jserr68594 | | #17

    Dana, I had one other thought. I have ducts in the floor of every room. What if I were to install a single head unit above the 2x4 window in the living/kitchen area and flow the air from that room into the other rooms via the floor/wall ducts? This would allow me to just go with a single zone minisplit. I could pipe the ducts together and add in duct fans that would suck air from the living area near the minisplit and blow it into the other rooms. Is this a good idea or bad idea?

    Attached in the pictures that shows where the ducts in the floor/wall are.

    Thanks again for all of your help and insight. I really appreciate you taking the time to type out detailed responses! You are a wealth of information!

  18. WashingtonChris | | #18

    Thank you all for your Questions and Answers... !

Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |