0 Helpful?

What size heat pump for a very small house?

I am getting different opinions on the correct sizing of a ductless mini split air-source heat pump. the mitsubishi rep recommends an 18,000 btu unit but the daiken rep thinks we can use a 12,000 btu unit. thoughts? it's for a 610 sq' house with 8' ceilings, r38 walls and r60 ceiling in -32 design temp. we will use electric baseboard heat for back-up. thanks!

Asked by erik olofsson
Posted Jan 11, 2013 2:46 PM ET
Edited Jan 11, 2013 3:37 PM ET


11 Answers

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Has anyone performed a Manual J calculation (or a similar heat load calculation) for your house? Such a calculation requires information on house orientation, window orientation, window sizes, glazing specs, and air leakage rates.

More information on Manual J calculations here: Saving Energy With Manual J and Manual D.

Even without doing a heat loss calculation, it's fair to say that your design heat load is likely to be under 12,000 Btuh (not Btu).

Here's the wrinkle: what is the outdoor temperature at which the output of the heat pumps was calculated? To the best of my knowledge, Mitsubishi Hyper-heat models produce their rated output at an outdoor temperature of 5 degrees F, and achieve 87% of rated output at -4 degrees F. Below that temperature, their output continues to drop.

Answered by Martin Holladay
Posted Jan 11, 2013 3:32 PM ET
Edited Jan 11, 2013 3:40 PM ET.


thanks again, martin

Answered by erik olofsson
Posted Jan 11, 2013 5:41 PM ET


-32*F is a brutally cold winter design temperature, so, absent data about windows and subslab insulation, I don't share Martin's confidence that heat load is below 12kBtuh.

Definitely do or get done a Man J load calculation. Though minisplit heat pump cold weather performance has greatly improved, I doubt any brand or model would have any significant heating capacity at -32*F. Therefore be sure backup heat is sized to meet entire load.

Answered by Curt Kinder
Posted Jan 11, 2013 11:03 PM ET



An R-2000 plan evaluator normally does a room by room modelling in HOT2000 before you even start construction and they should have provided the output of this modelling (including room by room and whole house heating and cooling loads AND all details/specs needed to meet the R-2000 energy target) to you and your HVAC contractor.


Answered by Gio Robson
Posted Jan 11, 2013 11:55 PM ET


Are the Daikin and Mitsubishi reps just throwing out numbers, or have they perhaps done calculations they're willing to share with you? What information have you provided to them?

Answered by David Meiland
Posted Jan 12, 2013 12:47 AM ET


david, i suspect the reps are guestimating. they asked about sq' and ceiling height and design temp. thats about it..

Answered by erik olofsson
Posted Jan 14, 2013 1:08 PM ET


The difference between the different sizing between the two vendors may be the lowest allowable operating temp, the output that temp, and the granularity of the sizing. Both Mitsubishi & Daikin cold-weather units will put out about 15,000BTU/hr per ton of cooling rating at an outside temp of -15C (+5F). From there it begins to taper off rapidly. While both Daikin & Mitsubishi publish output specs at -20C (-4F), only Mitsubishi has a spec at -25C (-13F), and that spec is about 10,000BTU/hr per ton of cooling rating.

So a 1.5 ton "Hyper Heating" Mitsubishi (the "...FE-18" ) will deliver ~15,000BTU/hr at -13F, and may be breaking even with resistance heating for efficiency at -25F (but delivering only 10-12KBTU/hr). But the Daikin may have already stopped breaking-even at -15F (or not- request an engineering spec for the low-temp limits, and the approximate output capacity curves.).

A house that size with those insulation specs would probably still have a heating design load under 10KBTU/hr @ -32F unless it has a lot of window area and low-performance windows, and 3/4-ton mini-split might still be the right choice. Specifying the mini-split for good performance at the binned hourly coldest month AVERAGE makes more sense than shooting for full coverage at the 99% condition, and oversizing it for the average load by more than ~50% will lead to lower comfort, and it would short-cycle during the shoulder seasons.

Answered by Dana Dorsett
Posted Jan 14, 2013 7:26 PM ET


Based on Erik's other posts, I'm pretty sure the design temp. is 32C, not 32F.

Answered by John Semmelhack
Posted Jan 14, 2013 8:55 PM ET


dana, i was hoping you would comment. john is correct, however, and the design temp is -32 celsius. the house has minimal north/east/west glazing and has 70 sq' of south glazing. the windows are all triple glazed fibreglass units made by duxton in winnipeg. it is pretty rare to get a day colder than -30c and we have allowed for electric baseboard heat in each room. i believe there are about 5600 hdd. thanks again, everyone...

Answered by erik olofsson
Posted Jan 15, 2013 1:13 PM ET


It doesn't make much difference, since -32C is only 6F/3.5C warmer than 32F, and the coldest temp at which any mini-split has a known-rated output is still -25C.

Got a location to look up on weatherspark.com, to estimate the winter mean temps, etc? (Annual HDD have no bearing on the sizing of the equipment.)

With limited glazed area and high-performance windows your heat loads are probably still well under the output of a 1-ton Mitsubishi (MSZ / MUZ-FE12NA )@ -25C. Assuming an interior design temp of 20C, exterior -32C (~94F delta):

If you have a total of 100 square feet of ~U0.22 window (they might be better than that), the window losses are only ~2000BTU/hr

With 610 square feet of ~R60 ceiling, the ceiling losses are only ~1000BTU/hr.

Assuming two ~20' insulated (R4) doors, that's another ~1000BTU/hr.

With R38 walls, that could be a whole-wall number, but let's assume some thermal bridging bringing the average down to R32, assuming you have a perimeter length of ~100' and 8' tall walls, that's 800 square feet of U0.03 wall, for something like 2300BTU/hr.

Add it all up and you're still under 6500BTU/hr at design temp, plus ventilation air.

The rated heating output of the -FE12NA is a bit lower than some 12,000BTU/hr-cooling mini-splits at 12,500BTU/hr @ -15C/+5F, but it's still putting out about 10,000BTU/hr @ -25C. see:


I'd be surprised if the 9000BTU/hr-cooling -FE09 wouldn't do just about as well, since it's capable of delivering 11,900BTU/hr @ -15C/+5F:


At -25C it'll be good about 75% of that 11,900 BTU/hr number, or ~8900-9000BTU/hr.

Do a REAL heat load calc on the place, using it's REAL U-values, but I suspect the 3/4 ton unit is going to be a far better choice than the 1.5 tonner.

Answered by Dana Dorsett
Posted Jan 15, 2013 4:57 PM ET


thank you so much for the analysis, dana. i can get both diakin and mitsubishi heat pumps in 12,000 btu units. do you recommend one manufacturer over the other?

Answered by erik olofsson
Posted Mar 15, 2013 5:14 PM ET

Other Questions in General questions

Difficulty finding a modulating multi-zone air-source heat pump

In Mechanicals | Asked by Jonathan Lawrence CZ 4A New Jersey | May 21, 18

Insulating Wall Between two Different Slab Elevations

In Green building techniques | Asked by Jonathan Lawrence CZ 4A New Jersey | May 24, 18

Vaulted / Cathedral Ceiling -- Unvented

In Green building techniques | Asked by User-7075735 | May 23, 18

Should I insulate and/or heat my unfinished basement?

In General questions | Asked by J Greene | May 24, 18

How do we hire a green building advisor on an hourly basis?

In GBA Pro help | Asked by Joe Martin | May 24, 18
Register for a free account and join the conversation

Get a free account and join the conversation!
Become a GBA PRO!