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

Adding minisplits to second floor of Cape Cod Style

aps171 | Posted in Mechanicals on

I recently had a HVAC contractor come out and give me some options on my cape cod style house in suburban Philadelphia (Zone 4A). Currently there are no return ducts from the second floor and minimal supply ducts as well, leading to a noticeable temperature difference on the second floor. The house layout does not allow new duct work to be run to the second floor easily, so it would be expensive and probably still undersized.

The house layout is as follows: The second floor consists of 2 bedrooms on either side (roughly 12’x16’ each) and a bathroom in the middle all connected by a hallway and the stairs. Total second floor area is 500sf. The walls are R-15 and the ceiling is R-50 in the middle and R-17 in the slopes. The first floor total area is 1000sf (2 bedrooms, living room, and kitchen). The basement has R-25 on the walls and R-5 above the slab and is 800sf. The house is about equal to current code minimum tightness based on a blower door test.

The house currently has 17 year old H/AC units: a natural gas furnace that is 80,000 BTU (80% AFUE) and a 2.5 ton AC. The HVAC contractor proposed adding mini-splits to the second floor and updating our furnace and AC for the first floor and basement as follows:

Basement and First Floor:
60,000 BTU furnace and 2.0 ton AC

Second Floor:
2-(1 in each bedroom) 9,000 BTU minisplit indoor units (Mitsubishi MSZ-GL09NA-U1)
Outdoor unit (Mitsubishi MXZ-2C20NAHZ)

The minisplits seem oversized to me but they did run Manual J calculations. They were afraid that putting just 1 unit on the second floor wouldn’t keep the other room comfortable as cool air would go down the stairs and hot air would get caught in the hallway. How does the community feel about this solution?

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. Expert Member
    Dana Dorsett | | #1

    It's not just the mini-splits- overall the proposed equipment looks MORE than 2x oversized for their actual loads. The target oversize factor would be in the 1.25-1.5x range, which is enough to cover you during a typical cold-snap, or even a major Polar Vortex event.

    Did they SHOW you their Manual-J calculations, or better yet provide you with a copy?

    I'm not quite sure what the minimum modulated output of the MSZ-GL09NA-U1 is when married to a multi-split, but when it's on a MUZ-GL09NA single-zone compressor it's minimum is 3600 BTU/hr in cooling mode, 4500 BTU/hr @ 47F in heating mode.

    I doubt the design heating load of the ~200 square foot bedrooms is more than 3500 BTU/hr @ +15F, unless you have single pane windows without storms, or it leaks a ton of air, and at +47F its probably half that or less.

    But the minimum modulation of the heads might not matter too much, the MXZ-2C20NAHZ multi-split has a 7400BTU/hr min output @ 47F. With a single head running it has to take the full output, and a pair running would have to split it, so even with both running they won't drop below 3700 BTU/hr @ +45F.

    A pair of separate MSZ/MUZ FH06s would would more efficient, but possibly more expensive too.

    Internet pricing on the MXZ-2C20NA + 2x GL09NA is about $2400 (just the compressor and heads), and a single FH06 mini-split run about $1500 each, or ~$3000 for a pair, and you'd have an additional compressor to mount, with associated hardware etc. It's probably a $1000 cost adder, but it'll be more comfortable and more efficient.

    A mini-ducted 3/4 ton Fujitsu probably would probably be about right, but it's not clear how easy it would be to install. With a bathroom and hallway between the two bedrooms there may be a way to do it with the mini air handler in a dropped ceiling in the bathroom or a the back wall/side of a closet something, but it's hard to make that call without being there. There would have be more construction involved, to be sure. The Fujitsu mini-duct cassettes can be mounted vertically or horizontally and have a bit more blower drive than the competition, and can throttle back to 3100 BTU/hr @ 47F, and could heat/cool the bathroom as well as the two bedrooms.

    The total capacity of either a Fujitsu 9RLF or a pair of FH06s is still likely to be overkill, but the lower minimum modulation keeps them running efficiently & comfortably despite the oversize factor.

    The cheapest thing to do for the second floor might be to just provide better return paths via jump ducts, etc and use the central system, but there may be floor-to-floor seasonal temperature balance issues if you go that route.

    A 60K furnace or 2 ton AC for a 1000' above-grade + insulated basement zone seems like extreme overkill. The heat load for the basement is probably no more than 3K, and the basement's cooling load is latent-only (unless you're planning to throw polka parties with a dozen beery stomping Czechs down there :-) ).

    The 1000' first floor MIGHT have a 1-ton of latent + sensible load, but not 2. At 2 tons you're at a ton per 500' of floor, which is the cooling load ratio of a barely insulated house in Louisiana with clear glass single panes!

    The heat load of the first floor @ +15F (Philly's 99% outside design temp) in an insulated sub- 3ACH/50 house is probably less than 15,000 BTU/hr. Add that to ~3K for the basement and you're at 18K, but it's probably less than that.

    You can probably heat/cool the first floor + basement with a 1-ton or 1.5 ton mini-ducted Fujitsu mini-split. The 1-tonner is good for a bit over 15,000 BTU/hr @ +15F

    The 1.5 tonner is good for over 20K @ +15F:

    The heat loss characteristics of basements vary quite a bit from fully above-grade floors with outdoor temps. In the basement the load is small and doesn't vary near as much with outdoor temperatures. As a result it's almost impossible to get reasonable floor-to-floor temperature balance between a basement and first floor. You dial in the flows at 40F outdoors and it may be quite a bit off at 25F outdoors. But since the basement's loads are tiny it may not really matter much- you can just open or close the register based on how it's faring on any given day, and don't bother tweaking the balancing vanes once it's anywhere close.

    If you have a heating history on the place, run a fuel-use heat load calculation on it using the 80% furnace as the measuring instrument on last winter's gas bills (winter bills only- shoulder seasons have much larger error.) If I had to guess (which I do, without more information) the real heat load of the entire house is probably no more than 25,000 BTU/hr @ +15F, and it may even be under 20K. It would at least give you a sanity check.

    As a point of reference, I live in a 1.5 story 2x4 framed 1920s bungalow with cellulose in the walls and an R15 insulated basement, but my roof-R is ~R20, and overall my R values are lower than yours. It's 2400', with 400' on the second floor, 2000' on the first floor, plus a ~1500' of basement. An IBR load calc on the place comes in at ~35K @ 70F indoors, +15F outdoors, ~40K @ +5F (my 99% outside design temp).

    A Manual J on my house would likely come in around 32K @ +15F. A prior Manual-J (at my cooler design temp of +5F) came in at about 37K, which hews pretty close to the fuel-use calculated load. (Within measurement error.)

    Your house almost can't be anywhere near as lossy as mine since it's higher-R. with a first floor HALF the size, basement a barely over half the size, and a comparably sized second floor. A 60K furnace wouldn't be extreme overkill at my house at +5F, but it's probably ~2x oversized for my load at +15F, and could be ~3x oversized for yours. With hot air furnaces oversizing that much still isn't an efficiency problem but it IS a comfort problem. But the AC oversizing factor is more likely to run into both efficiency & comfort issues.

    Paying a competent engineer or RESNET rater to run an aggressive load calculation will usually save you more in equipment costs than the fees charged for doing the Manual-J.

  2. aps171 | | #2

    Dana, thanks for your response and all the responses I have read by you on this site. I ran some of the calculations you provided and as you guessed came up with a much lower value than the HVAC contractor (only 11.5k BTU last January). I have asked the contractor to provide me with his Manual J calcs and I am waiting to hear back from him. He did say that his calculations used a 0 degree design temp heated to 70 degrees while my calcs used 15 to 65, but I ran those numbers in your calculations and still only got around 18k BTU.

  3. Expert Member
    Dana Dorsett | | #3

    Even though it might hit 0F in Phildephia during a Polar Vortex event, it's a ridiculous number to be designing a heat pump for, given the 99th percentile temperature bin is +15F. That all but guarantees the heat pump runs at a lower efficiency during the shoulder seasons and won't hit it's HSPF numbers.

    For a hot air furnace it's OK, as long as it's not up-sized very much from the load at 0F. With a hot air furnace shoot for something close to a 1.4x oversize factor for the load at +15F, but not much more . At a 1.4x oversizing factor for the load at +15F you'd be fully covered even at -7F.

    I suspect 0F is even beyond the 99.99th percentile temperature bin for that location, meaning fewer one hour out of every 10,000 hours is colder than 0F. It happens, but it's not every year, and it doesn't stay that cold for more than a very few hours at a time more than once per decade.

    The house has thermal mass- it has to stay colder than that for awhile before it gives up much ground even if you don't have quite enough capacity to keep it fully 70F, at night, with no lights or other "extra" heat sources going. After daybreak on a 0F night you have solar gains to help out, and rising air temperatures.

    There are 8760 hours in a year. If you design to the 99th percentile load you're completely covered for all but 87.6 hours out of the year. If you're sized EXACTLY for the 99% load there may be a night or two in a given where you'd need an extra blanket or some auxilliary space heating, but one step-size up in equipment usually has quite a bit of margin.

  4. aps171 | | #4

    They HVAC people got back to me with an updated Manual J calculation. All it contains is the summary so I don't know exactly what values they used for different things. It came to 47,000 BTU/hr heating and 26,000 BTU/hr cooling. The design temperatures are more correct on this one too. Talking with them I know they assumed my house to be very loose just because of the age and construction type. I reminded them about the blower door test showing that it was relatively tight for what it is though. I'm not sure how much of a difference that makes, certainly not the only driver.

    I used on my house using very conservative numbers and got 24,000 BTU/hr heating and 20,000 BTU/hr cooling.

    Right now their proposal is for a 3 ton heat pump (2 stage) with a 60,000 BTU (97%) backup furnace. No minisplits will be added at this time. They did say they would be willing to put in a 40,000 BTU furnace although that is not their recommendation. I could drop to a 2 ton heat pump for cooling but I'm not sure if it's ok to stick with a 3 ton in order to gain the extra heating capacity. A 3 ton seems like it would work down to at least 10 degrees and probably even as low as 5 degrees, while the 2 ton would only work to around 20 degrees, not factoring in the difference in cost between electric/gas. They also said a 5 stage heat pump probably would not work great for my house because of the poor existing duct work, although I intend to fix some of that myself.

  5. Expert Member
    Dana Dorsett | | #5

    Most Manual-J software reports the amount of load attributed to infiltration, and an assumption of "leaky" can be a very large number. If they won't share that number with you or any of the input assumptions it's not clear you can really work with them. (You also have an insulated basement, which takes down the heat load by quite a bit.) A 40K furnace is overkill, a 60 K furnace is extreme overkill.

    A whole house heating load of 24,000 BTU/hr @ +15F for 1500' of above grade conditioned space above an 800' conditioned basement is a credible number. That's 16 BTU/ft of fully conditioned space, which is probably on the high side of reality. But let's assume it's real: The ASHRAE standard would be to install a furnace with a 1.4x oversize factor, which would be 33,600 BTU/hr-out, which would be ~35K-in for a condensing furnace. A 40K furnace wouldn't be terrible- just less than ideal. A 60K furnace would be terrible.

    In reality the 24K number is probably a bit padded, and 2-stage multi-speed 30K condensing furnace (eg: Goodman GMEC960302BN ) would be just about right, if a gas furnace is the way you want to go here. Even at low-fire most 2-stage 40K furnaces would be delivering more heat than your design-day load. A 60K 2-stager would put out almost 2x your design day load even at low fire (enough to cover the load at nearly -40F outdoors even at low fire) which is simply ridiculous. See page 3:

    A whole house cooling load of 20,000 BTU/hr for 1500' of conditioned space is also a very credible number.

    If I had to bet on whose numbers were closer to reality it's really no contest- you win. There's no way your heat load is anywhere near 47,000 BTU/hr @ +15F with the windows and doors closed. It's probably a bit less than half that.

    A third party Manual-J with a full report performed by a competent professional who lives and dies by the accuracy of their numbers will save you more in up-front equipment than the fees charged.

    In situations like this you may be stuck figuring out yourself the optimal equipment to install, and putting it out to competitive bid. (Been there, done that...) I'm still thinking a mini-ducted 1.5 ton Fujitsu for the first floor (with ducts in the basement) and a mini-ducted 3/4 tonner (or a pair of Mitsubishi FH06s) for the upstairs would have you covered down to sub-zero temps at much higher comfort levels than anything the contractor has come up with.

    Alternatively, an 1.5 ton Mitsubishi PVA-A18AA7 bigger-deal air handler and a pair of FH06s on a 3-zone 2ton or 2.5 ton multi-split compressor would likely work:

    The full sized Mitsibishi air handlers have a 3:1 turn down ratio, which is a bit more modulation range that you get out of Carrier Greenspeed or similar.

    Don't let the HVAC contractor size it for you- get a third party room by room Manual-J, and be willing to pay something for it. Even if it's a grand (probably half that) it'll be worth it.

  6. aps171 | | #6

    Dana, thank you for your responses. I did your heat load calc on the coldest 6 days the past 2 years and the highest use I got was just over 17,000 btu/hr so your estimates seem to be right on. I agree with the heater size. What do you think about the heat pump? Would a 3 ton be useful enough for heating (paired with solar panels) to be worth the oversizing of the cooling aspect. Honestly we don’t use our ac unless it’s at least approaching the 90s so I don’t care about the cooling as much as the heating. Thanks again.

  7. Expert Member
    Dana Dorsett | | #7

    A 1 or 2 speed 3 ton heat pump is extreme overkill for a heat load of 17,000 BTU/hr @ +17F, and wouldn't meet it's HSPF numbers.

    And 3 tons (36,000 BTU/hr) of cooling is way overkill for a calculated cooling load of 20,000 BTU/hr. Why would you want to oversize it by almost 2x if you don't care about cooling?

    A 2-ton Mitsubishi multi-split with a 1-ton or 1.5 ton real air handler and a pair of FH06s would still have plenty of max-cooling capacity, and better latent load handling, and would pretty much hit it's HSPF & SEER numbers. So would a pair of mini-ducted Fujitsus.

  8. aps171 | | #8

    Dana, thanks again for all this great information here and in all your other posts. I will look into everything you said.

  9. Expert Member
    Dana Dorsett | | #9

    BTW: "I did your heat load calc on the coldest 6 days the past 2 years and the highest use I got was just over 17,000 btu/hr..."

    I'm not sure how you ran a the fuel use based heat heat load for a random selection of cold days(?). Accuracy isn't very good over very short periods. Over longer periods the BTU per degree-hour constant can be inferred with reasonable accuracy from the fuel used and the heating degree-day data from a nearby weather station. Deriving it from fuel used in a single day or a selection of 6 random days has a lot of data "noise" in it.

    If you were just using the 6 coldest daily lows over a 2 year period as the outdoor design temperature and applying the BTU per degree-hour constant to it that would introduce some error too, but it the error wouldn't be as large. Ideally you'd use the 99th percentile temperature bin for all hours of the past 20-25 years as the outside design temperature.

  10. aps171 | | #10

    Yeah, I figured those results would give me lots of noise, but I basically took some cold few days and applied the same formula and then averaged it out. I figured that would be basically be giving me a worst case scenario as the temps were around (maybe slightly below for the low, like +9 degrees compared to +14 design temp at worst, and right around design temp for the high on those days). Then I averaged them out and added a little margin for the amount of fuel used since it's only in whole digit CCF/day. We have had pretty mild winters here the past 2 years so I didn't want to underestimate my use over a full month. This made the fuel use change from 11k btu/hr for the full month average to around 17k btu/hr for the coldest days average. Really this was more of a sanity check then anything and seems to line up not too bad my load calcs. I also couldn't go back more than 2 years because the insulation and air sealing work I have done would not have been included.

  11. Expert Member
    Dana Dorsett | | #11

    An aggressive Manual-J would likely confirm your fuel use numbers if done over a longer period, but it's the right order of magnitude, and certainly nothing like the 47,000 BTU/hr the contractor came up with. (How do they even get away with such sloppy BS? They're counting on the customer's ignorance being as deep as theirs, I suppose.)

    Using the "before upgrades" fuel us numbers over longer periods would also be a good sanity check, and is one way of measuring just hoe much the upgrades have improved things.

  12. aps171 | | #12

    Good idea on the old one, maybe I'll check them just to get a feel. I will say this is the only company that even came in with the 60k. All the others wanted to do an 80k since that's what I have now even though the efficiency would be going from 80% to 97%! It was all I could do to not laugh out loud when I saw that.

  13. Expert Member
    Dana Dorsett | | #13

    Join the club!

    Maybe there should be an "Oversizers Anonymous" 12 step program for HVAC pros?

    There are way too many 80 & 100,000 BTU/hr furnaces heating houses with <25,000 BTU/hr loads in the US. With hot air furnaces 3-5x oversizing is the rule, not the exception, in part because it's not a big cost adder up front, and it doesn't negatively impact efficiency, and they never EVER have to calculate a heat load to be sure it can heat the house. They get used to seeing a certain size and just assume it's "normal" without questioning whether it's NECESSARY.

    The price difference between a 40,000 BTU/hr furnace (a 1.6x oversize factor for a 25K load) and a 120,000 BTU/hr furnace (~5x oversize factor) within a manufacturer's series is usually less then $500. The difference in cost for a condensing single stage 40K and the 80K version is less than $150, so who cares? Why take the chance that the load might be more than 40,000, if simply doubling the output costs the contractor far less than the cost of running a Manual-J?

    Tract house builders typically spec the same furnace for the whole development, even when the house size & loads vary by 2/1, and it's 2x oversized even for the house with the largest load. They get quantity discount pricing and the installers don't have to adjust the learning curve for size since they're all the same.

    But oversizing by a large fact DOES negatively impact comfort. For an industry that purports to be primarily selling comfort the practitioners don't seem to be getting the message on how to make that happen.

    If taking that same approach with heat pumps efficiency suffers right along with comfort (by quite a bit), and the pain of the up-front wallet-ectomy for the upsizing is more severe.

    But when it comes to equipment replacement time it's the homeowner, not the home builder who is in the driver's seat. A homeowner keyed into the nature of the problem is more willing to spend a few hundred on the Manual-J (and save a few hundred or more on equipment costs) than an HVAC contractor or home builder.

Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |