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

Sizing the minisplit: ducted upstairs, ductless downstairs

johnchard | Posted in GBA Pro Help on


Firstly thanks for the all the fantastic info you continue provide on HVAC sizing. The information asymmetry between the contractors and the average homeowner is incredible. We are looking at retaining an energy consultants to run manual J and or manual D calcs and I am sure what we outline will reinforce this for you,

We have just purchased a 1910ish Colonial in CT (Zone 5). Single family dwelling two stories with 3 beds and 1 bathroom upstairs and a fairly open plan downstairs. The house has no central ac and uses a forced air oil burner for heat. The home size is 1225 sq ft and was renovated in last ten years- plaster walls; double pane windows, new foam in ceiling in attic floor. There is ductwork in floor downstairs and in plaster ceiling upstairs via unvented attic.

The standard approach seems to have been a walk through before sending us a quote for Mitsu systems. The $ values of the quotes have ranged from $16K to $30K. We rec’d a fourth quote in the $24K range but lost the precise mechanical breakdown.



Heat Pump

Air Handler (upstairs)

Wall Units (downstairs)


Vendor 1



MSZ-GL12NA  * 2 units


Vendor 2





Vendor 3





My CoolCalc numbers (note: I have zero experience doing this) gave me Heating BTU downstairs of 10,951 and upstairs of 25,771

1. I am guessing this is nuts oversizing? 
2.Does ducted upstairs and ductless downstairs make sense/right approach? I wasn’t keen to use the old ductwork downstairs.
3. Does 1 register/ductless wall unit downstairs make sense – all contractors initially pushed two on us
4. Per the “Getting the right mini split” article on this site, should we be looking at a Fujitsu over a Mitsu for a house this small?
5. Is a multi split a bad idea ie we do separate outdoor units for upstairs and downstairs?

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. user-1072251 | | #1

    Have you had a a blower door test or a heat loss analysis? Minisplits work best in a tight house, and while your house has some spray foam, that does not mean it is a tight house. Before spending money on a heating system, learn how your house works. And old fashioned hot air ducts are not suitable for minisplits anyway; the ductwork for them will be shorter and of a specific size. We installed three wall miniplits a few years ago in our old 2600sf partially upgraded house; best heating/cooling/dehumidification system ever, but the gaps show up in midwinter. (Our cost:$11,000)

  2. MattJF | | #2

    Post a copy of your coolcalc output. This seems a little high, but not wildly. The load difference between upstairs and downstairs is seems a bit bigger than expected. What are the wall/ceiling R values? Coolcalc is can be buggy, you need to check the ventilation and infiltration loads and that the areas are computed correctly. is another one to try. For heating loads, you don't need to worry about the orientation categories and can plug the areas into a single line per wall type.

    If it is going to be awhile before you do this upgrade, get an ecobee or other logging thermostat. You can use the log data to do a very accurate heating load calculation.

    Does the oil furnace serve both floors?

    It is often possible to use much of the existing duct work with a ducted mini split. You would need to check the total static pressure the furnace is operating at and what CFM it is moving from the furnace fan chart. A ducted minisplit should move much less CFM, and produce a much lower pressure loss. Use the following equation to get an idea of what the static pressure will be for a ducted minisplit on the same ducts. You can get a cheap digital manometer on Amazon.

    Sn = So(CFMn/CFMo)^2

    Sn = New Static Pressure
    So = Old Static Pressure
    CFMn = New CFM
    CFMo = Old CFM

    You will need new return and supply plenums to mate with the old duct work. The old duct work should also be sealed if it is not currently. Watch out for duct runs in exterior walls as they can have condensation issues, particularly if running a minisplit in dry mode.

    Two systems is generally a better option for efficiency. Fujitsu is often the go to for slim duct options around here. They make a standard unit that does .36 inH2O static or a mid pressure unit that does .72 inH2O static.

  3. Jon_R | | #3

    Be sure to also estimate your load from your old heating bills.

    Do some blower testing/sealing if it hasn't been done previously.

  4. Expert Member
    Dana Dorsett | | #4

    >"Post a copy of your coolcalc output. This seems a little high, but not wildly. "

    The Coolcalc load numbers sound about right for a leaky house that has no wall insulation and wood sashed single panes. Seriously- ~11KBTU/hr downstairs, ~26K upstairs, for ~36-37K total for a 1225' house, with outside design temps in mid to high single digits?

    That's 30BTU/hr per square foot, which is about TWICE the ratio to expect at a CT type design temperatures for a TIGHT 2x4 framed house with code-min low-E replacement windows. If the foundation isn't insulated it might end up around 20BTU/hr per square foot. Most homes in 1910 didn't have crazy high window/floor ratios, unlike some mid-century modern style homes with high load per square foot ratios.

    It's definitely high by more than a little, and edging into "wildly" territory. A better description of the house and it's location (got a ZIP code?) might help.

    If the prior owner was on a regular fill-up service that stamped a "K-factor" on the fill up slips it's possible to estimate the whole house load from wintertime oil bills, as Jon R was suggesting.

    If the actual load is more likeky 20-25,000 BTU/hr the MXZ-4C36NAHZ is about 2x oversized, and sub optimal, independently of which cassettes/heads are being installed. That thing can deliver 45,000 BTU/hr @ +5F, which is pretty close to (and probably cooler than) the 99% outside design temp. See:!/product/25915

    In fact, most multi-splits will have a minimum modulation on the compressor that's pretty high for loads as low as I'm expecting.

    1. johnchard | | #5

      Dana (and Jon, Matt and Bob),

      Thanks for all the contributions and thoughts.

      1. We will contact Eversource for a blower door test. Unfortunately Energy Vanguard won't take on another client due to their current workload. Welcome other suggestions.
      2. Oil furnace served both floors. The previous owners said heat averaged around $140/$150 per month and electricity around $220 per month. This includes an electric resistance water heater from 1999 rated at 5,358 kw/hr per yr. (don't start me on this... seems like there isn't a plumber in CT who wants to install a Sanden for me). This was for 2 adults and 3 kids. We are looking at PV before end of year.
      3. Coolcalc attached.
      4. House is 1916. House had a major refurb at some point. All gaps seem to have some spraying of foam. Basement the rafters have glass batts but rim joists are not insulated. Obviously the house is leaking but at some point an effort was made to tighten it up. Plaster work on inside is great. I dont know what is under the vinyl siding. Attic at some point had cellulose before they removed it and foamed floor of attic. I would like to use Havelock batts and Intello Plus in attic rafters.
      5. Windows are some double hung (I think thats what are they called) double pane vinyl windows. Not a lot of windows in this house and they are small.
      6. Zip code is 06820

      Thanks again.

      1. johnchard | | #6

        Additional images

  5. Expert Member
    Dana Dorsett | | #7

    >"Oil furnace served both floors. The previous owners said heat averaged around $140/$150 per month "

    Over the past couple of years the average #2 oil price has been about $3/gallon:

    So $150/month would be $150/$3/gallon- 50 gallons/month. At 138,000 BTU/gallon that's 6.9 million BTU (MMBTU)/month, source fuel energy. At 85% combustion efficiency the net that went into the heating system is then 5.865 MMBTU.

    From the monthly average temp numbers on the electric bill it looks like the colder months run about 32F. A full day at a steady 32F would be (65F - 32F=) 33 heating degree-days (base 65F). A 30 day month of 33 heating degree-days would be 990 HDD.

    So... 5.865 MMBTU / 990 HDD= 5924 BTU/HDD, or in a 24 hour day (/24=) 247 BTU/degree-hour.

    The design temperature of +9F is (65F - 9F=) 54F heating degrees below the presumptive 65F heating/cooling balance point, for an implied load of about...

    54F x 247 BTU/degree-hour= 13,338 BTU/hr

    ...which would be a pretty tight well insulated 2x4 framed 1225' house, or they used deeper than deep setbacks.

    If they were getting $2/gallon oil (in their dreams) instead of $3 oil raises the load to 13,338 BTU/hr x $3/$2= 20,007 BTU/hr.

    Or may be the outright lied to you about how much they were spending on oil?

    But 20,007 BTU/hr / 1225' = ~16 BTU/hr per square foot, a credible number.

    13,338 BTU/hr / 1225' = ~11 BTU/hr per square foot- on the low -load side for a 2x4 framed house @ +9F, but not insanely low.

    In the CoolCalc 198 square feet of window for 404' of floor area upstairs is an insanely high 49% window/floor ratio. Most new houses are 15% ratio, most 1910 antiques are closer to 10-12%. Is this the turret of a lighthouse? :-)

    The window heat loss of 10,508 BTU/hr for 198 square feet of window is about 53 BTU/hr per square foot, at an indoor to outdoor temperature difference of (70F - 9F=) 61F degrees. That's a U-factor of (53 / 61= ) 0.87 BTU/hr per degree-F, which is roughly the performance tight wood-sashed SINGLE pane, no storm windows. A current code-minimum double pane window would come in at U0.32.

    So something is way off hear, and I haven't really started to dig into it. There is probably other stuff off in other parts too. There is no way to verify the "Construction Nr" types without the CoolCalc tool, so you'll have to go over it bit by bit to remove as much garbage-in = garbage-out type problems.

    1. johnchard | | #8


      Many thanks again.

      Oil usage - I have no way of confirming this. Its a pretty old furnace - all of the hvac contractors called it a 'Cadillac' and one went so far to say he estimated it running at 65 percent efficiency. The oil per month jumped out to me knowing how much our family nearby spends on heating oil for their place.

      Windows - adjusted to double pane

      Infiltration changed from tight to loose

      Added kitchen mudroom to downstairs space (not sure if this makes sense)

      Result - downstairs at 15k btu and upstairs at 19.6K. Windows proportionally in range you mentioned.

      Given EnergyVanguard passed on us - anyone else who could help us? How far can I get with CoolCalc and the GBA/Community help?


      1. bfw577 | | #9

        I am in CT as well. Go to and schedule an energy audit. Its like $149 and they will do up to 1k in work. They will do a blower door test and duct leak test.

      2. Expert Member
        Dana Dorsett | | #10

        >Windows - adjusted to double pane

        Is that the only option? No low-E? Sometimes it's more accurate with some of these low-end tools to select "Triple Pane" if they don't have a low-E double pane window type.

        >"Infiltration changed from tight to loose"

        It should be entered as TIGHT rather than loose. The defaults for infiltration in most of these tools are ridiculous, but I've never played around with CoolCalc for comparison.

        Even though it doesn't make pretty printable graphs, does at least an OK Manual-J, if you enter "tight" under "construction type" and are as aggressive as possible on the R-value options. Even then it still tends to overshoot by 25-40% but would be fine if you DON'T oversize the capacity of equipment from their numbers.

        >"...anyone else who could help us?"

        This outfit in Monroe CT claim to do it by the book, on-guard against oversizing, designing systems for architects & builders:

        These folks based in Chester CT (a bit further afield, to be sure) do HERS / RESNET ratings and other levels of analysis, and can probably do a straight-ahead honest Manual-J (and other things):

        This company in Waterbury does Manual-J load calculations as a service:

        They also install stuff, so watch out for a thumb on the scale, and impress on them that you are more concerned about keeping the oversize factor down to less than 1.3x than you are about getting cold.

        I ran a search for a RESNET/HERS rater within 100 miles of Darien on this site:

        I'm sure some of these people will do Manual-Js as a service:

  6. Expert Member
    Dana Dorsett | | #11

    Looking at the floor plan you have six (possibly 7, counting the bathroom) windows on the upstairs. Yet the CoolCalc is calling out 198 square feet of window. That would be 33 square feet per window, which is ENORMOUS for a 1910 colonial.

    Your down stairs CoolCalc indicates 96 square feet of window for 8 (?) windows, a more realistic 12 square feet average per window.

    Most homes of that vintage have something like 30" x 58" = 1740 square inches which is (/144=) ~12 square feet, or 32" x 60" windows for 1920 square inches/ 13.3 square feet. Even at 15 square feet per window you'd only have 90 square feet, not 198 square feet.

    In recent memory, I've only seen one antique house that had a few 15+ square foot windows. That was an 1840s house with visually huge ~40" x 75" windows (about 21 square feet per) in some of the first floor rooms.

    As it happens that house is currently on the market so there are some web-pics. Are your upstairs windows 50% bigger than these?

    There are mid-century modern houses with some crazy-sized windows with crazy-sized window/floor are ratios, but I'm pretty sure your house doesn't look like that, but it would have to for the area indicated in the CoolCalc. Your real window area on that floor is probably about 75 square feet, 80 tops. Measure again. Your whole house may not even have 198' of window area, let alone JUST the second floor.

    Measure up the wall area carefully too, subtracting out the areas of windows & doors.

    Regarding the walls...

    5810 BTU/hr for 982 square feet of wall is 5.9 BTU/hr per square foot at a delta-T of 61F, for a U-factor of 0.097 BTU/hr per square foot per degree-F. That implies something like R11 cavity insulation with a 25% framing fraction half-inch wallboard and half-inch plywood sheathing with wood clapboards or something.

    Most homes in 1910 were built with full-dimension 2x4s, which if blown full of cellulose comes in at R15-ish at center cavity. The framing fraction also increases with full dimension lumber but so does the R-value of the framing fraction, and if balloon framed it'll still be lower percentage than the current 25% typ. Plaster on wood lath is higher R than half-inch wallboard, and 0.75-1" wood plank sheathing is higher R than half-inch CDX. You're probably looking at something like ~U0.085 for those walls depending on the actual material stackup, insulation, & framing type. If so it would mean the load per square foot of wall is overstated by about 14%.

    The 5654 BTU/hr of duct loss is from what? Are the ducts outside the insulation in an attic or something?

    1. johnchard | | #13



      New CoolCalc attached. I think I am getting the hang of this. Have tried to measure and tweak as much as possible. Window sizes have been adjusted and changed to low-e double pane (except the old door in family room and the back door). Infiltration changed to tight.

      Overall the BTU's for the house have come down. The addition of the windows in the downstairs bathroom and mudroom have added to BTUs. Overall window area landing where you estimated.

      On the question of the oil - the former owner confirmed that amount per month. She said it was a fixed monthly plan at around $140 per month. She ran the night time thermometer at 61 and daytime temp of 65.

      Duct loss upstairs is because the air handler will be located upstairs in attic. The original current & uninsulated duct work is in the attic. I will insulate the attic. and add a vapor retarder at some point.

      Loadcalc - Ill try this over the weekend.

      Energy Audit - thank you. Will try EdgertonHVAC.


    2. johnchard | | #14

      Dana and everyone, thanks for the help.

      1. I have Eversource contractor doing a blower door test tomorrow
      2. Thanks to Steven Winter Associates initially helping, I now have Tony Listanti at Integral Building and Design Inc. sorting out the analysis.

      Will come back with end findings.



  7. walta100 | | #12

    I see this bid as the first step of the bait and switch sales tactic.

    Step one give the customer the mini split system bid they asked for but inflate the bid by planning a head in every room along with over sized compressors.
    Step two acted shocked when they call to see why you have not returned the signed the minisplit bid.
    Step three present you with a bid for a much lower priced conventional system. Note the conventional system will have ½ the capacity of the proposed mini system and it will still be twice as large as necessary.


Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |