Retrofit single ductless minisplit to cool entire upstairs
Hello, I’m a huge fan of this website, but I finally joined so that I could ask this specific question:
My wife and I own a 2000 sq-ft colonial built in 1962 in the Metro Detroit area (climate zone 5), with only baseboard heating–no central air. The first floor can be kept comfortable with a single wall-mounted AC unit and some fans to help circulate, but the upstairs is a bigger challenge. We’ve had several options quoted to provide some form of AC, but we’re leaning strongly towards a ductless minisplit system.
However, the problem is that I don’t know what arrangement makes the most sense for the upstairs. The main challenge is the layout; it’s essentially a 16 foot hallway with 4 bedrooms off of it. Originally we were quoted for a 4-zone system, with a head in each bedroom. However, after further reading, it seems that design would be overkill… especially after getting a cooling load calculation of roughly 10,000 Btu/hr for the entire upstairs! (Though even that # may be high… the heating load calc they provided was about 70% higher than reality!)
I’m wondering, could get away with installing a single 9K Btu unit in the hallway, and have it service the entire floor? There are several articles that talk about using a single unit per floor, but they’re regarding “tight” houses. Our house is likely drafty by GBA standards, but last winter we had a major sealing and insulation job done on the house… basically made the attic, basement rim joists, and overhangs air-tight, and put cellulose insulation (R-49 worth) in the attic. The CFM-50 went from 3645 to 2465.
So the question is, how well would a single 9K minisplit be able to cover the upstairs of our house? My intention would be to leave it running all summer, in order to keep the whole floor roughly 72-75F and dehumidified. Also, we would leave bedroom doors open during the day. My hope is that we wouldn’t need additional fans to push the air into bedrooms. Would there be enough cooling through convection and the internal walls to combat the heat entering through the external walls & windows? Each bedroom has 2 small double-paned windows, and we get enough shade to prevent much direct sunlight, except for early morning and late evening.
Thanks in advance, and sorry for the long post!
GBA Detail Library
A collection of one thousand construction details organized by climate and house part
This is a tough question to answer. I think that the main variable is the expectation of your family members. If you are all flexible, and are willing to put up with some room-to-room temperature variations, I think that a ductless minisplit in the hall (with open doors during the day) would work fine.
If you suspect that not all family members will be flexible, you might look for a closet on the second floor where you could install a ducted minisplit near the ceiling. Then you might be able to install ducts from this unit to various bedrooms (assuming that your ceiling is high enough to box in the ducts).
With the doors open during the solar-gain hours convection should do pretty well at keeping the room temperatures bounded (though they won't necessarily be at the same temperature.) If it's the standard wall coil type, think about which direction it blows relative to the door openings or you may end up overcooling some rooms, and under cooling others. Also consider where the stair well is located, since there will also be convective air motion happening there. With a layout of the floor plan to ponder the head placement options might become clearer.
In some cases you'll be better off with a floor-mounted version rather than a wall-coil, if hallway width or ceiling heights interfere with a wall-mount.
No matter what your heating fuels & system is, it's worth the upcharge for a full heat pump version rather than a cooling-only mini-split, to provide both backup auxiliary heating, as well as low-cost (usually cheaper than condensing natural gas) heating during the shoulder seasons when the heat loads are low (say, when it's in the mid-40s or warmer outside.)
Hi Mr. Holladay, thank you for your prompt reply!
I believe the family will be pretty flexible, considering the current lack of AC! However, I would like to strike a good balance between flexibility, comfort, and energy use if possible, given the constraints of the layout. We do have a hall closet that is currently unused, however the upstairs ceiling is pretty low (maybe 7.5'), not really tall enough to box ducts in under the attic space.
If we do use a single head in the hallway, what would the expected temperature difference be from the hall compared to the individual rooms? Would we have a 70F hallway but 76F bedrooms? We have ceiling fans in the bedrooms, would they help in any way to mix/pull some of the cool air from the hallway? (I know normally they shouldn't be run when nobody is around to benefit from the breeze)
Conversely, if we did a multi-zone setup with a head in the three used bedrooms (1 is spare/office), would it just be problematic in other ways? The load for the master bedroom is 2.2K (3K if including attached bath), but the other two bedrooms are only 1.5K and 1.8K. Even if all doors are open and the units can each cool 1/3 of the upstairs, at most each head is only responsible for 3.3K of cooling (on the absolute hottest days), but the smallest heads are designed for 6K.
I feel like the upstairs load is small enough to warrant only a single head, however the distribution / airflow is the main problem... and I don't have a good grasp on how that works. Unfortunately, none of the 4-5 AC sales people I've met with seem to be building science experts, which is why I'm asking here! :)
With a closed door, I don't think you will be happy with bedroom temperature or humidity. Even when moderately over-cooling the hallway.
I don't advise the three-head approach.
I'm not going to predict temperatures. (And in any case, sunlight through a window will change everything.) In my mind, there is nothing wrong with a hallway that is a little cool if the result is bedrooms that are comfortable.
Hi Jon, I agree that having a closed door during the day would make a bedroom uncomfortable, but I'm willing to tell the family that they can choose between a cool bedroom or privacy (at least during the day). We have two preschool-aged children so daytime privacy won't be an issue for at least 10 years (I hope!).
Mr. Dorsett, thanks for your response as well! The upstairs layout is basically one 16' hallway, with stairs on the east end, 3 bedrooms on the south side and a master bedroom on the north side. I'll include a rough sketch of the floor plan. The head wouldn't blow directly into any particular bedroom. However, it would be blowing directly into the stairwell leading down. Would that be a problem (ex. stratification?), assuming we keep the downstairs at a similar temperature? FYI, planning to install a 12K unit for the main floor to replace the existing wall AC (and make it more centralized).
I'd definitely like to get the heat pump versions, but mostly because of the increased SEER ratings. It may also find use as a backup in case our boiler goes out... but probably won't use it during the shoulder seasons because that's when our new condensing boiler really shines! Also, Michigan's electricity sources are still not very green--though neither is fracking...
Excellent, thank you for all of the input!
It takes a lot of sun-exposed window to keep a bedroom 10F warmer than the adjacent space if the door is open. If the doors are open until a few hours after sunset, the temperature difference shrinks pretty rapidly. With no direct sun, even at 87F (Detroit's 1% outside design temp) the cooling load attributed to conducted heat through the walls is pretty small, and even a 5F difference in temperature drives quite a bit of open-door convection.
With open doors the humidity (which plays a large factor in overall comfort in a Detroit summer) is going to be evened-out, even if temperatures are not. Unless you are spending time in the bedrooms with the doors closed during daylight hours when it's north of 85F it's not likely to become a comfort issue, even if the peak delta-T is 10F on a PM sun-drenched west facing bedroom.
A room with a design load under 3K can't be efficiently served by it's own ductless head, since the the minimum modulated output of the head is above the average load, and it will do a lot of on/off cycling at minimum speed rather than modulating with load. When operating in that mode it's latent cooling falls off too, leaving the humidity higher than with a single head in the hall serving up the whole-house load. IMHO you will probably be HAPPIER with the overall comfort range with the hall-mounted head than with individual room heads, as long as the doors remain open until after dark, when the direct wall heating and window gains go away.
With a tight/very-tight house very little latent load from infiltration is picked up overnight. If a bedroom seems to start out comfortably dry when you close the door go to bed, but you wake up sticky a few hours later (even without much of a temperature change), it 's probably an indication that the room could use more air sealing effort.
As drawn that should work pretty well, especially if Bed 1 and the bathrooms have no west facing windows.
Bath 2 would get almost no sensible cooling, but with the doors open it would get at least some latent cooling (that could be helped along with a bathroom exhaust fan, if there is one.)
Bed 3 probably has less load per square foot that the others, since is has no south or west facing windows, but it's temperature may end up a bit warmer than Bed 2 and Bed 1 due to the orientation of the air flow relative to the door. It will get convection air exchange, but very little direct-blow to help it along.
The shoulder seasons is also when the mini-splits really shine. Not just shine but really REALLY shine, actually beating their HSPF numbers (which are based on the seasonal average), as long as there is enough load that it's mostly modulating. A 3/4 ton mini-split with an HSPF of 13 means when reasonably sized for the load has seasonal average coefficient of performance (COP) of 3.8. But when it's 45F outside and operating at a lower speed it's COP beats that average by quite a bit, and 2x the efficiency when it's +17F outside and running nearly full-tilt. Take a look at Figure 5 on page 10 of this document:
This older 1-ton has a tested-labeled HSPF of only 12.0 (= COP 3.5), which is roughly it's performance at mid-speed at 35F outdoors. But at lower speed at 47F it's COP is north of 5, and even at max-speed it's delivering 3.5. At 55F at low speed we can infer that it's COP would be running north of 6. Even that older mini-split would be averaging better than 5 during the 45F & warmer shoulder seasons, and a better-class current version would be averaging about 5.5. What does that really mean?
Running at a COP of 5.5 means it delivers (3.412 BTU/watt-hour x 5.5 x 1000 watt-hours/kwh=) 18,766 BTU of heat into the house per kwh of electricity used. Normalizing that to chunks of million BTU (MMBTU) delivered, it takes 1,000,000/18,766= 53.3 kwh/ MMBTU.
At at a typical MI residential retail rate of 15.4 cents/kwh that's ~$8.20/MMBU.
A condensing boiler running 95% efficiency delivers 95,000 BTU/therm , so per MMBTU of heat delivered it takes 1,000,000= 10.5 therms, but it also chews through some pumping & control power (varies quite a bit system to system).
At typical MI residential retail rate of ~80 cents/ therm (fully delivered price), that's $8.40/MMBTU,
If those are roughly your rates (take the April bills, divide the total $ by the kwh & therms to find out), it means heating with the condensing boiler runs a few % more expensive during the shoulder seasons than heating with the mini-splits. But that isn't discounting for the distribution heat losses (which mini-splits don't have) or the electrical power used by the heating system (which is already included in the mini-split's COP.) In reality it's going to be more than 10% cheaper at those utility rates.
During the coldest months the efficiency of the boiler drops off much more slowly than with mini-splits- and when it's +17F and running nearly full-tilt even a new cold climate mini-split's COP will be no better than 2.5. Under those conditions using the utility rates above, heating with the mini-split would be north of $15/MMBTU while the condensing boiler wouldn't be much more than $9/MMBTU even if running temperatures above the condensing zone.
So, that's why it's worth the up-charge for a heat pump version, and why paying attention to HSPF may be as important as the SEER numbers.
Also note: Ultra-high SEER usually means low latent cooling when operating at low speed. Most mini-splits have "DRY" or "DEHUMIDIFY" cooling mode to help purge moisture when it's sticky out, but it's as-used SEER when operating that mode is dramatically lower than the tested SEER. In Detroit summertime outdoor dew points are pretty high, and you will likely end up using those modes at least half the time to keep the indoor humidity at comfortable levels.
[ This was heavily re-edited to better reflect MI utility rates- I had somehow used Chicago rather than Detroit on my first go-around. I must need more PM coffee or something... :-) It's a far more favorable in Chicago than Detroit, at state-average rates, but the actual rates at YOUR utilities matter.]
If you don't find a single head working well when doors are closed, consider pressurizing the house to the point where air is flowing from the hall to the bedrooms, not the reverse. Otherwise you are attempting radiant cooling with a leaky (probably) building, upstairs (ie, normally infiltrating) and a humid climate.
Holy cow, thanks for chugging through all the numbers! I actually used your Carbon Footprint article to get a rough estimate while deciding if it made sense to completely replace my existing baseboard heating with heat pumps. When I ran the numbers, it looked like a boiler upgrade would be cleaner to operate over a whole season, at least until 2030, due to Michigan's rather weak energy targets. But that was also based on an HSPF of 10.2 (I don't recall which product I was considering at that time). I figured by 2030, both systems might need replacing anyway, so I could re-evaluate then.
Also, in terms of carbon footprint, I thought I read that electricity has like a 60% efficiency from the power plant to what actually reaches your home (obviously that might be mitigated with solar panels). But that might be a misunderstanding on my part.
OK, back to the topic at hand... The heat pumps I am looking at--primarily for AC purposes--are a Mitsubishi FH09NA for upstairs and FH12NA for down. I liked those models for the wide modulation ranges and SEER ratings, but could they have trouble with latent cooling? I was assuming that if I left them on all day (with a moderate temp setting), then they'd continually suck the moisture out of the air while running at a lower speed (i.e. more efficiently).
FWIW, here are the calculated loads for the house:
Cooling (Design = 93F outdoor, 72F indoor)
Upstairs = 10595
Main = 15078
Latent = 3463
Heating (Design = 3F outdoor, 72F indoor)
Upstairs = 17486
Main = 24350
Basement = 17173
However, we'd be fine setting the thermostat closer to 75 on a really hot day, so those numbers would be smaller. Also, the heat load calc must have been extremely conservative, because the total load for the house was 59K, but in reality on an 8-degree day, the boiler was keeping our house 70F with roughly 32K output! I imagine on design day, it'd only need about 37K output tops. That means the calc was about 60% too conservative. So I don't know if the cooling load calc used similar false assumptions.
Edit: if/when I do get the heat pumps installed, I will definitely try out a shoulder season to see how well they perform for heating. I didn't realize how crazy efficient they can be for the lighter loads.
Edit 2: I should mention that both full baths have 110CFM fans installed, which I think will help some with the latent load. Also, we'll be installing a 450CFM range hood in the kitchen (though I doubt it'll ever be run that high).
The FH09 is good for 12,000 BTU/hr at the standard 15F temperature difference, with an indoor relative humidity is 51F (=dry bulb 80F, wet bulb 67F.) From the submittal sheet:
Cooling | Indoor: 80º F(27º C)DB / 67º F(19º C)WB; Outdoor: 95º F(35º C)DB / 75º F(24º C)WB*
Even if you drop the indoor design temp to 75F you're looking at an 18F temperature difference, and a somewhat higher latent load than test conditions (unless you can tolerate 66% RH). It'll probably still keep up at 75F, most of the time (and keep up with the latent load too), but it's guaranteed to be running at maximum speed at least some of the time, and may not always be able make your desired temperature during peak gain hours.
Max cooling for the FH12 at a 15F delta-T is 13,600 BTU/hr, so it'll probably lose a bit of ground at the 1% outside design temp, assuming the load numbers are accurate, but probably not enough to really matter.
If they're not quite covering the humidity when idling along at low speed on muggy non-torrid days, run them in DRY mode until the sensible load picks up a bit.
Ok... from the sounds of it, 9K upstairs and 12K downstairs will be a bit undersized. Would it make more sense to bump them up to the next sizes, i.e. 12K and 15K respectively? From what I understand, the heat pumps are more efficient when a little oversized, since they hit peak performance closer to 50% output. But I'm clueless as to how latent and sensible heat are controlled properly. When it comes to comfort, I'd much rather have warmer air but lower RH.
Assuming the load calcs are accurate, what size heat pumps would you recommend?
Also, I should mention that there is one east-facing window in bed 2, and one west-facing window in both bed 1 and bath 1. So that might really heat things up during the peak part of the day. 9K may not cut it...?
Part of my problem is, I'm not sure how accurate the cooling load calc was... the same company included heating load calc, and it ended up being way higher than what I actually experienced last winter (59K design day vs. 32K when it was ~5 deg warmer).
Thanks again for all of your help! If I'm being too much of a drain on your time, please let me know (or feel free to ignore the remainder of this thread). I would try to get this sort of info from the AC contractors but so far only 2 of 6 have even done load calcs! And they either push central air ducted in unconditioned attic, or 1-head per room.
Sorry to resurrect this old thread... but I have a question about something Mr. Holladay said in his original answer. He mentioned installing a unit in a closet and running ducts to the bedrooms. If I am looking at Mitsubishi products still, would it be something like the MVZ-A12AA4 air handler? Or is there a certain indoor unit that acts more like the wall-hung units but can be ducted somehow? In the picture I added earlier, I am considering putting something in the closet that's centered on "Spare Bed". It's roughly 26" wide and 24" deep, and opens to the hallway.
In order to reach that closet, we'd either need to run the coolant & drain lines through the unconditioned attic (bad!) or along the ceiling and frame it in. I am not opposed to either option, but I'm wondering how much of an efficiency hit the system would take in the attic; I've read that it's roughly 20-25% by having ducts up there, but not sure if it'd be as terrible for the supply... especially if it can be dropped between joists for the whole run, and buried in the loose cellulose insulation.
Similarly, how awful would it be to run North/South portions of the duct system between the joists in the attic, to reach the far sides of the rooms? For example, I was thinking I could have framed-in ducts running along the south wall in the three bedrooms, but I'd need to either run a dropped duct across the hallway to reach the master bedroom, or run it between joists.
The MVZ-A12AA4 air handler is a fairly beefy sucker, suitable for driving a whole-house duct system.
The SEZ /SUZ series mini-duct cassettes are a lot cheaper/smaller, and that can work as long as the duct runs are fairly short & hard-piped (no flex duct.) eg:
If the duct runs are very long getting the air flow can be more problematic- the Mitsubishi SEZ mini-duct cassettes are on the wimpy side compared to the Fujitsu RLFCD series competition:
Both series come in 3/4 ton & 1.5 ton versions too, and Daikin has a similar line up, if it turns out you have better local support for Daikin than the others.
Running the refrigerant lines through an attic is no big deal, but running condensate drains or ducts in the attic would be bad. It's better to build out soffits below the ceiling level for the duct runs if you have the head room.
The Fujitsu models seem better in every way... maybe I need to talk to a authorized dealer? :) I see at least 10 in my area.
How long is too long, when it comes to these mini-duct cassettes? If the cassette is centralized, it would be roughly a 14' run for Bed 3, 10' to Spare Bed, and an additional 5' each for branching to Beds 1 & 2. I know practically nothing about ducts, so I'm not even sure how feasible my idea it would be. I've attached an updated sketch with ducts instead of ductless.
Our ceilings are only 7.5', so we have very little room to spare for soffits--maybe 6" max, and that may not be enough to hide ducts in. One thing I'm considering is, having the main N/S branch in the attic, between the joists, and having it spay-foamed and buried in cellulose. Then possibly, the E/W branches could be placed in a smaller soffit on the back wall, above the windows.
My design would also need to run the supply & condensate lines in the attic between the joists. Would condensation be an issue if these were also hit with closed-cell foam to seal in the moisture?
Thanks for all of your help. I'm hoping to talk to some HVAC guys soon to go over design ideas but a lot of times they don't have high-efficiency in mind.
Since MI summers tend to get muggy, good humidity control is a concern for me. Is there an easy way to tell how well these ducted units would perform? I see the Fujitsu 9K removes 1.6 pints/h, and 12K removes 2.7. Some Mitsubishi submittals mention it but others don't. The FH09 wall unit is 0.6 pints/h. Is that why you mentioned needing to run in DRY mode on cooler humid days?
Should I be looking at SHF or the removal rate, or both?
The 9RLS3 is really more like a 1-ton by some measures. It has a minimum modulation nearly twice that of the FH09, and 22% more heating capacity. Either one should cover your loads just fine, but the FH09 will modulate more and run more continuously.
Either one will will probably need to be run in DRY / DEHUMIDIFY mode at outdoor dew points hit the high 60s if there isn't much sensible load.
The only time the mini-split would actually need to remove more than 0.6 pints per hour would be if the house is actively taking on wind-driven rain through an open window. That's a LOT of moisture, more than would be brought in at normal ventilation rates.
Regarding heating efficiency for shoulder seasons, should I look at the manufacturer's COP @ 47F numbers, or try to estimate based on HSPF? For the 12RLS mentioned in that study, I am guessing it has a rated COP of 3.91 at 47F, since the 12RLS2 is rated as such, and also has a HSPF of 12.
I noticed that sometimes the HSPF and [email protected] don't correspond well... for example, one unit has HSPF and COP = 11.7 and 3.3 respectively, but another has 10.3 and 3.83. Also, the Hyper Heating units seem to have better [email protected] than their regular counterparts.
If looking at just the shoulder seasons, the COP @ 47F or is going to be more relevant than the COP @ +17F, but neither is going to be super-relevant if it's doing a lot of cycling during the shoulder seasons.
HSPF is a bit messy, but as I understand it it's based on COP performance at the "rated" or "nominal" heating performance at both +47F and +17F, a presumed oversizing factor relative to the heat load @ +17F, and a presumed zone IV number of heating season hours in this map:
Technically Detroit is in zone V on that map but with modulating systems the real oversize factor relative the real heat load,. and the temperature at which it begins to cycle on/off skews real world performance by quite a bit (in both directions.
HSPF is supposed to represent some sort of seasonal average performance in a zone IV location when sized reasonably for the heat load, but it's possible to hit or beat the numbers in a zone V location if the minimum output is low enough that it still doesn't cycle much even when somewhat oversized.