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

Replacing a natural gas gravity furnace in an upgraded small older home

gworrel | Posted in Mechanicals on

I am a regular reader here and have learned a lot on this site. Thanks to all who contribute.

I have been upgrading the thermal envelope in a 960 square foot house in Detroit, Michigan for several years. I have injected polyurethane foam in the walls, R60 cellulose in the attic after air sealing, 2″ XPS on the basement walls, and 2″ XPS in the joist cavities in the basement. The windows are single pane but with caulked in place storm windows.

I have done a heat loss calculation using the page at using R values from Using a 65% efficiency rating for the existing natural gas gravity furnace, I can get the annual heating cost to match up with reality. The calculated heat loss is 14,300 btu/hr at a design temperature of 6 F. The 6 F was chosen based on the 97.5% temp for Detroit.

My annual heating costs were about $400 this last winter which was a much worse than average winter.

The house has a 768 square foot conditioned basement that has a home theater area and a workshop so total conditioned space is about 1728 square feet. I have a wood burning stove in the main living room but I don’t use it and may remove it entirely. The main reason for getting rid of the gravity furnace is to get rid of the large ducts. I really don’t want any ductwork in the basement if at all possible.

I was originally thinking about radiant pex in the basement joist spaces, but that is costly and complicated. One of Martin’s posts here started me looking at wall mounted natural gas furnaces. The Rinnai direct vent furnaces look interesting. They are priced between about $700 for the 8000btu model to about $1250 for the 17000 btu model.

I am thinking I could put in two of the smaller Rinnai units, 8K or 11K BTU, one in the basement and one on the main floor. The basement could be kept cooler when I am not using it and turned on as needed and in very cold weather to add to the total heating of the house. An 11K unit on the main floor and an 8K unit for the basement would be about $1650. Seems tough to beat for a new heating system. Any comments?

Right now I use window air conditioners but would like to put in one or two mini-splits at some point mostly due to noise issues. The high expense of the low temperature units makes me think I might be better off with one or two wall mount gas furnaces (one for basement and one for main floor), and one or two mini splits for cooling only.

I have not done a separate cooling calculation. Is there a good similar spreadsheet for that? Something like a Pioneer 12k mini split is only $525 at

How would I calculate the payback on a higher SEER model?

Any feedback would be appreciated. Thanks.

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. davidmeiland | | #1

    Just curious, what is the fixed monthly charge for gas service, and can you get rid of it entirely if you switch to electric for everything?

  2. gworrel | | #2

    The monthly service charge is about $10. I have a natural gas water heater which I plan to replace with a direct vent unit. The chimney is old and I'd rather not use it at all. I also have a swimming pool heater that I fire up once or twice a month in the summer for pool parties,so turning the gas off entirely is not a good option. My gas rates are about .90 per ccf and electricity is about $.167 per Kwh.

  3. GBA Editor
    Martin Holladay | | #3

    Since it sounds as if your house is relatively tight and well insulated, I think that your plan to heat the house using gas-fired space heaters with through-the-wall venting will work.

    However, if you really are planning to install a few ductless minisplit units in the future, I strongly urge you to simply install minisplits that will work for heating as well as cooling. This approach is likely to cost less than gas space heaters plus minisplits.

    For more information on these issues, see:

    Heating a Tight, Well-Insulated House

    Choosing HVAC Equipment for an Energy-Efficient Home

    Heating Options for a Small Home

  4. Expert Member
    Dana Dorsett | | #4

    +1 on Martin's mini-split recommendations. If you are going to be cooling with mini-splits it's probably going to be comparable or cheaper up front to pay the up-charge for a heating/cooling heat pump mini-splits than do deal with the gas plumbing to multiple wall-furnace units. If your total heat load is ~15KBTU/hr @ +6F a pair of 3/4 ton cold-climate mini-splits (or a single 1-ton on the main floor, with radiant cove heaters as auxilliary heat in the basement if it isn't going be used much.) will cover you with room to spare. Models to be looking at would be:


    Mitsubishi MSZ-FH09NA or MSZ-FE09NA

    Fujitsu AOU-9RLS2, or AOU-9RLS2-H


    Mitsubishi MSZ-FH12NA or MSZ-FE12NA

    Fujitsu AOU-12RLS2, or AOU-12RLS2-H

    Just be sure to install the outdoor unit where it's protected from snow-clogging from drifting snow or roof avalanches. Bracked mounting it on the wall above the snow depth line, protected by the overhanging rake of a gabled roof works. (Under eaves is OK, but rake is better.)

  5. gworrel | | #5

    Thanks for the replies. I just spent the morning running some more numbers. The cost of electricity here is $.146 per Kwh for each additional. To get that I just looked at several bills and calculated the incremental usage versus incremental cost. The cost of natural gas is about .82 per 100ccf. I used the same procedure and got varying results from about .79 to .84 so averaged to .82.

    Comparing the Fujitsu 12rls2h to the Rinnai efficiency of 82%, using about 30 million BTU annual heating load, the annual cost is about $300 for the Rinnai, and $471 for the Fujitsu. I used this calculator here: I used 272% efficiency for the Fujitsu. Of course that doesn't include the electrical costs of the Rinnai. heaters. One reason the Rinnai units caught my eye was the modulating gas valve and variable speed blower which should make them more efficient and quieter.

    By those calculations, the mini split could cost more for heating than what I am spending now. I could easily have a mistake in my calculations. It looks like the Fujitsu unit is probably quieter than the wall mount gas furnaces. Tough to find any dB numbers on the gas units.

    I started using a portable A/C in the basement last year, primarily for dehumidification. A stand-alone dehumidifier seemed to heat up the basement. Today, I am not running any A/c and it is 82 degrees outside and the same temp in my home office. In the basement it is only 73 degrees. So a mini split in the basement is probably overkill.

    The simplest solution would be a single mini split on the main floor and supplemental heat in the basement. But would it cost me more than heating with gas? I could continue to use the portable A/C for cooling in the basement. The basement is used about 2-3 hours a day in the winter, at the most.

  6. davidmeiland | | #6

    At those utility rates and equipment efficiencies, gas is much cheaper heat. Gas would have to move to $1.25 per therm to equal mini-split electric, and if it did that, I would expect electric to go up too, since a lot of it comes from NG.

    Still, you have low loads, so you might find that the added cost of installing the gas equipment makes it cheaper overall to just use the mini-splits, even if the monthly utility cost is a bit higher.

  7. Expert Member
    Dana Dorsett | | #7

    "I used 272% efficiency for the Fujitsu."

    That is an unrealistically low estimate for the efficiency of that unit for a Detroit climate. In fact you would be average about 300%, maybe a hair better. The 12RLS2 beat the 1-ton FE-series Mitsubishi in third party bench testing, and fleet of ten FE12s seasonally averaged a COP of 2.96 in field testing at Idaho Falls ID, which averages about 5F cooler in winter than Detroit. The 12RLS2-H has an HSPF of only 9.3 , compared to the 12RLS2 (12.0) or the FE12NA (10.6)

    And with the -FH12NA (HSPF= 12.5) you'd be bumping on (or even exceeding) 350%- it's more efficient than the 12RLS2, and WAY more efficient (34% more!) than the 12RLS2-H.

    So if it costs $471 to heat with the 12RLS2-H (probably won't, but just sayin') it would be about $350 with the -FH12NA

    Gas is currently at near-historical lows, but it won't stay there forever. Electricity is going to see some price inflation over the next decade, but is likely to see some price deflation after 2025 due to the anticipated ridiculously low cost of photovoltaic solar.

    So even if you would end up paying $50-150 more per year to heat with a mini-split this year than with the Rinnai, it's likely to even-out (or even flip) within the lifespan of the equipment. And the cooling efficiency of a high efficiency mini-split means you use about half or less power for cooling than you would with a best-in-class window air conditioner (and it's a heluva lot quieter.)

    The Rinnai gas units are pretty quiet. In my experience I'd say they're comparable to a 1.5 ton -FE18 at mid or high-speed, not as quiet as the RLS2-H series or the FH-series Mitsubishis.

  8. gworrel | | #8

    Thanks Dana. I took another look at the heat loss and ran the numbers at I came up with 18954 Btu/hour, not subtracting heat gain. Cooling load is 9190 Btu/hour. (Is it standard to adjust cooling load for bodies but not heat load? That is how his calculator works. Add people and cooling load goes up but heat load stays the same.) Since I subtracted heat gain from the load earlier, the comparable number is 17706. I can get that to jibe with my actual heating costs if I assume an efficiency of 80% for the gravity furnace. The only numbers I can find about gravity furnace efficiency say anywhere from 50% to 65%. The low efficiency numbers would probably make sense if I was not heating the basement. The basement contributes about 6000 Btu/hour to the heat loss.

    I have some room to improve the heat loss numbers at the back of my house where there is a shed roof addition which held the original bathroom and kitchen. I am doing some renovating back there this summer and expect to improve the insulation.

    I found that Rinnai does rate their units for noise at 31-38 dB, which puts them right about where you put them.

    You have me thinking more about forgetting about the Rinnai gas units. If the cost of operation is close then it is tough to justify the capital expense and installation for the gas units in addition to the mini splits. If I use a 12KBtu mini split on the main floor and a 9K Btu unit in the basement, I should be well served for both heating and cooling. If the basement is kept cooler except when in use, the cost of heating could easily go down. Whether that is practical depends on how quickly a mini split can raise the temp 10 degrees or whatever the difference might be. The radiant cove heaters look interesting for areas of the house that may run a bit cold.

    What I'll need to do now is run some cost and payback numbers on the different units to compare initial cost and operation cost over the next 5-10 years. The Mitsubishi FH12NA series looks to be the most efficient but also perhaps the most costly. With the design temp of 5F (99%), do I need to worry about the performance at -13F? We did get that cold here last winter but that was unusual.

    Panasonic and Gree are two brands that seem worth considering. The Gree 12K Terra series (25 SEER, HSPF of 9.0) was on sale at Home Depot for 1049 in May. Not sure if they might still do that price. Now it is $1399. (12-24 months same as cash on my HD card is attractive.) It is available for $1080 from AC wholesalers. The Mitsubishi is $1800.

  9. Expert Member
    Dana Dorsett | | #9

    The FH12 would probably run just north of $4K to install in my neighborhood (specutive, since the distributors in my neighborhood don't have them yet, and only released the wholesale price lists to the installers last week, but it probably won't be a huge upcharge over the -FE12)

    You don't really care about the performance 99.99% outside temperature bin (which is probably about -10F) , the 99% bin is the right temperature to design to. On the coldest day of the decade you can leave some lights on or fire up a space heater, or close some doors and stay in the warmer room if it's losing too much ground. But you DO need to have a unit with a specified output that meets the load at +5F, your 99% number. (The Fujitsu AOU-12RLS2 without the -H is both more efficient and cheaper than the AOU-12RLS2H, and nearly as efficient in heating mode as the -FH12NA.)

    Many units work great for air conditioning, but fall off a capacity-cliff below +15F or so- read the specs carefully.

  10. gworrel | | #10

    I wanted to compare different mini splits that are frequently recommended here and also a couple of lower priced units to see which ones offer the best price/ performance mix. So I put together a spreadsheet that uses manufacturer data to project annual costs for heating and cooling based on 40 million BTUs a year to heat and 350 hours of air conditioning.

    I am primarily interested in relative measures rather than actual expected costs here, but I did use my best guess for my house and climate. I used a fairly inexpensive ($644 internet price) mini split as a baseline (Gree Rio 9K) and compared all the rest to it.

    I was a bit surprised at the result. The least expensive Gree unit came in dead last. The new Mitsubishi 9K FH09NA came in first in lowest 10 year costs, including initial outlay. The Mitsubishi was $3350 cheaper over 10 years even though it was about $1000 more up front. I did not include any installation charges because I plan to do a diy install with perhaps some help with the final setup.

    Two others that stand out in the comparison are the Fujitsu 9RLS2 and the Panasonic XE9PKUA. The heating capacity of the 9RLS2 exceeds that from many 12K units. My design temp is 5F so I don't need the super low temp heating except rarely, but I included the -5F capacity if I could find it in the manufacturer's literature.

    I am no expert at this stuff so my calculations may be off. If any of the numbers look suspect, please let me know. I would also be interested in any other units that might be worth putting into the mix. I have attached a jpeg of the spreadsheet.

  11. Expert Member
    Dana Dorsett | | #11

    The numbers aren't perfect, but the gist is clear. I'm not surprised that the -FH series wins the 10 year cost of ownership, or that the -xxRLS2s score pretty high compared to the Gree & Panasonic units. The three-year cost of ownership will probably have a different stack up.

    The as-used HSPF isn't always comparable- it depends on both climate and oversizing factors. You're not likely to hit a true annualized COP of 3.95 for the FH09 in your application, but it's going to be over 3.5.

    Similarly you won't actually hit 3.5 with the -12RLS2 or -9RLS2, but it'll be over 3.0.

    Without explicit modeling of each model over temperature, you can estimate using clues in the max output at +17F, vs. output at which it was rated for efficiency at +17F.

    For instance, the Gree Terra submittal page ( ) shows a max capacity of 9350 BTU/hr @ +17F, and and delivers a COP of 2.7 @ +17F when modulated down to 8330 BTU/hr. Since +17F is below your winter average temp but above your 99% design temp, if sized correctly you'll be moduating a bit lower than that at +17F, and probably beat 2.7 for a seasonal average, which is already better than the 2.64 entered in your spreadsheet.

    Similarly, they rated the -FH09NA at +17F at a modulated output of- 6700BTU/hr, which is barely more than half it's max output at that temp, but it's using 600watts at that 6700BTU/hr, which at an input of 3.412 BTU per watts (x 600 watts) is a COP of 3.55. At it's max output of 12,200BTU/hr @ +17F it's drawing 1440 watts, x 3.412 BTU/ watt is a COP of 2.48. If your load at +17F is somewhere in between, it's COP will be in-between- call it 3-ish, but since your winter average temp is quite a bit warmer than +17F, sized correctly to modulate for most of the season you'll clear 3.5 for a seasonal average, but it's highly unlikely you'd be bumping on 4 since for most of January you'll probably only average about 3.

    For a DIY install, understand that you will have no warranty, and the tech who does the final install owes you nothing other than the hours & materials paid for. In snowy climates it's best practice to bracket-mount the outdoor unit on the wall above the maximum snow-drift line, and protected from roof avalanches/cornice-falls by some roof overhang (the rake being preferable to eaves), or a small shed roof. If you ground mount it or place it where it can be buried/clogged by snow coming off the roof you spend a lot more time digging out and de-frosting than you would otherwise. But even if you're happy to void the warranty, before buying a mini-split, scope out the amount of local distributor & installer support there is in your area- the last thing you want is to be stuck waiting 12 weeks for a repair part to be shipped from the factory in Taiwan or Korea to the sole distributor in California for reshipment to you. Both Mitsubishi & Fujitsu have pretty good (but not perfect) supply chains & support for their heat pump products in the US, not so sure about Gree or Panasonic. In a climate as cold as yours this can be a pretty important factor.

    The -xxRLS2 s are pretty good units, but they don't auto defrost, and defrost-water can contribute to ice buildup in the bottom pan high enough to damage the blades during weather that stays well below freezing for very extended periods. This is an issue for many mini-splits. The -H versions solve that by brute force with a pan heater that operates during defrost, which may account for it's sharply lower HSPF numbers. That said, there are many of them installed and running fine in places that hit -20F- just keep an eye on the ice build up issue during extended cold snaps until you're sure it's not going to happen to you.

  12. halcali | | #12

    are the fujitsu rls2s that much better at heating output than the mitsy fhs at cold temps (i.e. 14300 vs 9300)?

  13. Expert Member
    Dana Dorsett | | #13

    A sample size of only one, but here its the 3rd party bench testing the older -FE12 head to head against a -12RLS2:

    The efficiency of the -FH12 is higher than the -FE, but the capacity specs are nearly identical.

    Figure 4 is the capacity plot against temp for the 12RLS2. Figure 8 is the capacity plot against temp for the FE12.

Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |