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
Guest Blogs

Installing a Ductless Minisplit System

We decided to get rid of our oil boiler and install an air-source heat pump from Japan

Goodby oil boiler — hello minisplit. We are able to heat our Massachusetts house with a simple ductless minisplit system. Copper tubing and an electrical wire connect this outdoor unit with a fan/coil unit on our living room wall.

Image Credit: Marc Rosenbaum
View Gallery 6 images
Goodby oil boiler — hello minisplit. We are able to heat our Massachusetts house with a simple ductless minisplit system. Copper tubing and an electrical wire connect this outdoor unit with a fan/coil unit on our living room wall.

Image Credit: Marc Rosenbaum
We located the indoor unit of our new Fujitsu ductless heat pump high up on our living room wall.
Image Credit: Marc Rosenbaum
Dick and Tim Mavro came and took the oil heating system away. On the trailer you can see the boiler shorn of its blue Buderus jacket. The trailer also holds our old oil fuel tank and the indirect tank that the boiler heated for our hot water, as well as a 55-gallon drum with the remaining oil that we pumped out of the tank.
Image Credit: Marc Rosenbaum
I patched the flue pipe hole with rigid polyiso and plywood. What really struck me during this process was how dirty the burning of fuel oil is. The flue pipe was half blocked with soot (it should be cleaned annually, but because it is sealed no one wants to pull it apart and then re-seal it) and the boiler had about an inch of soot inside. And of course the oil itself is a smelly liquid hydrocarbon that we risk spilling in our homes and creating our own little Superfund site.
Image Credit: Marc Rosenbaum
As the oil tank fill pipe and oil tank vent pipe were removed, I notice that they slid right out, letting me know that there wasn't any caulk or other air sealing products around them. For now, the holes are temporarily filled with foam sealant.
Image Credit: Marc Rosenbaum
Vineyard Propane came by and took away my propane tank, making our house fossil-fuel-free at last. I recently sold our last remaining fossil fuel burner, the Vermont Castings gas heater. I'm supposed to get a refund for the propane in the tank, which had about 80 gallons according to the tank gauge.
Image Credit: Marc Rosenbaum

The Island Cohousing houses were designed to have heat and domestic hot water (DHW) supplied by an oil-fired boiler. (Time for a pedantic distinction: a furnace heats air and blows it around a house, and a boiler heats water which is pumped around the house).

They chose a pretty good boiler: a German Buderus G115. The two-bedroom houses got two heating zones’ worth of fin-tube baseboard heat, one zone per floor level. The three- and four-bedroom houses have a third zone, for the first floor ell.

When this boiler was specified, it used the smallest oil burner nozzle available, rated at 0.5 gallon/hour, and therefore its heating output was about 70,000 BTU/hour. (For geeks: this is higher than you’d think, because the nozzle is rated for an oil pressure of 100 psi, and the burner on this boiler is running at 150 psi, which means it bumps the nozzle flow up to about 0.6 gallon/hour.)

A small house in a mild climate

I’ve previously noted that these aren’t superinsulated houses, but they’re relatively small houses — 1,166 square feet to 1,800 square feet — and are located in a mild climate. So the design load of our house is about 22,000 BTU/hour. So on the coldest night of the year, the boiler only needs to run about 1/3 of the time. On an average winter day, it may run about half of that, and in the shoulder seasons, even less.

When the boiler is not firing, it loses heat up the chimney flue and off the boiler itself as it cools down. Our boilers here are sidewall-vented (don’t do this with oil, it’s a black mess on your exterior wall around the vent) and so may not be losing as much as if they might if vented vertically up a chimney — but here I speculate.

The boiler, DHW tank, and the uninsulated heating piping all reside in the basement.

On Martha’s Vineyard it is common to put fiberglass batts in the basement ceiling (the first-floor framing) and leave the basement walls themselves uninsulated. A basement like this can take all the heat loss you can throw at it. Heat goes into the ground and into the outdoors above grade. An 8-inch-thick concrete wall with an air film on the inside has a thermal resistance (R-value) of about 1.5, which is halfway between a single-glazed and a double-glazed window.

Right around Christmas, the basement was 59°F, which feels fairly temperate. Just before Christmas, we installed a Fujitsu 12RLS single-zone minisplit heat pump in our living area. (More on that in a bit.)

The boiler is keeping our basement warm

On the morning of December 26th, we turned the thermostats controlling the oil boiler down to the minimum setting so the heat pump could do all the heating work, and then the boiler only fired to make DHW.

I put a Hobo state data-logger on the oil burner. This nifty instrument shows when the burner is on and when it is off — it logs its “state.”

When the boiler was operating to make heat and DHW, I saw it run between 3 and 6+ hours daily. When the boiler operated only to make DHW, its operating time dropped to about one hour per day.

In what I recall to be little over a week, the basement temperature dropped about 10°F, to below 50°F. The difference was all the heat not being lost from the boiler and the heat distribution piping. Brrrr.

Monitoring the oil boiler run-time

We operated the boiler without using the second-floor heating zone and relied on the heat from the first floor rising by natural convection up the stairs to the bedrooms and bathroom upstairs. These rooms ran 2-4°F cooler than the main level.

For a week in early February, I turned the heat pump off and used the oil boiler again, so I could get some good data on burner run-time and oil usage. During this week it averaged about freezing outdoors, and from my burner run-time I figured we used just over 2.5 gallons/day of fuel oil. Some of this was for DHW, but less than would be indicated by the roughly 0.6 gallon/day we measured when the boiler only ran for DHW.

The reason is that when making only DHW, the boiler would run for about 20+ minutes every 8 to 9 hours, because it had cooled off in the interval and first had to heat its own water storage and cast-iron mass up to temperature before it could effectively heat the DHW.

Once the DHW tank thermostat was satisfied, the oil burner turned off, stranding all that heat in the boiler, there to cool off to the flue and the basement. When the boiler was making heat also, it never cooled off for very long, so it didn’t use as much energy to make the DHW.

Interpreting the data

The image below is a snapshot of the burner state from 6:00 a.m. one morning to 6:00 a.m. the next morning. The boiler fires pretty often, on for a couple of minutes then off for 6 or 7 minutes.

At about 3:00 p.m. and again about 10 hours later, there is a long cycle, about 13 minutes, for DHW. A little after 11:00 a.m., the boiler stops firing for heat — most likely a sunny day.

It picks up again around 6:00 p.m., then it’s off a bit after 8:00 p.m. It looks like we turned the thermostat down, because the boiler doesn’t fire again for heat until the wee hours.

During this week, we used somewhat over 2 gallons of oil daily for heat. That is perhaps 290,000 BTU per day input energy. (N.B.: Disclaimer — I’m really fond of Marc’s Sloppy Mathâ„¢. I don’t care much about 2% accuracy. Most useful conclusions and design directions can be made with 10% accuracy. I am good about getting the decimal point in the correct location, which comes from using a slide-rule back in the Stone Age.)

A new ductless minisplit

Now let’s compare with the energy used by the minisplit heat pump.

I’ll say more about the technology in another post, but the short story is that the Fujitsu is an air-source heat pump, which uses electrical energy to operate a compressor to extract heat from the outdoor air and deliver it to the inside of the house.

It takes no special genius to get heat to go from hot to cold, but the other way around is harder. Your refrigerator is a heat pump, using electricity to take heat from inside the cold refrigerator compartment and dump it into your warmer house.

The Fujitsu heat pump delivers warm air from a unit high on the the living wall. The hunky looking white pipe in the photo below contains the refrigerant lines and wire that connect the indoor and outdoor units. (Normally this would be outside but the gutter interfered…)

Ductless minisplits are energy misers

We’ve heated the house (with the exception of the ell, which we keep closed off from the house, and which seems to run at basement-like temperatures without direct heating) for a bit over three months now with the heat pump.

When the heat pump was installed, I had the electrician wire an old-fashioned electromechanical electric meter in line with the heat pump so I could see how much energy it was using. I’ve been reading the meter at least once daily.

On the coldest day, we used almost 21 kWh. On mild days more recently, we use as little as 3 or 4 kWh. During similar temperatures as occurred during the week we ran the oil boiler, the minisplit has used 12-13 kWh/day. Converting that to BTU gives a daily input of around 43,000 BTU.

The oil boiler was using 290,000 BTU, remember? How can the heat pump be over 6 1/2 times better?

Well, for one thing, the reason we use heat pumps is that they use one unit of electrical energy to move more than one unit of “free” energy in the outdoor air. This ratio is the Coefficient of Performance (COP), and higher is better. The Fujitsu is probably operating at a COP of 3 or more. So the energy input into the house might be 130,000 BTU or even more per day. Plus, the heat pump is not heating the basement or losing energy up a flue.

So we’ve gone from a cost of over $8 per day for heat, to $2.25 per day. And we’re heating to slightly higher temperatures, for reasons that need to be left to another post!

Marc Rosenbaum is director of engineering at South Mountain Company on the island of Martha’s Vineyard in Massachusetts. He writes a blog called Thriving on Low Carbon.


  1. Expert Member
    ARMANDO COBO | | #1

    Mark, according to MW Dictionary, an ELL is a unit of measure, or an extension at right angles to the length of a building. What are you calling an ell? An enclosed porch? Never heard that word before.

  2. GBA Editor
    Martin Holladay | | #2

    Response to Armando Cobo
    Armando Cobo,
    The "first-floor ell" is a room at right angles to the main rectangle of the house. Imagine the short leg of a capital "L" -- that's the room that Marc is talking about.

  3. user-984364 | | #3

    So, the basement?
    What's the plan to keep the basement above 50F, now?

  4. watercop | | #4

    Great Data!
    Would you kindly dig out of your data hat an estimate for the efficiency of the oil fired boiler while it is just producing DHW. I seem to recall your DHW use is quite low.

    I'm curious as to Energy Factor for a boiler just making DHW...I'll bet it is miserably low, perhaps in the range of 0.10 to 0.25

    10% accuracy with a correctly placed decimal point would be just fine!

  5. markgimmeshelter | | #5

    Another interesting blog Marc, Are you accounting for DHW in your bottom line?

  6. rkthomas13 | | #6

    missing units
    Anybody know where I can buy the airhandlers for an out of date mitsubishi mr. slim heat pump model PUH36EK. They are PCH9 or 12 GK.
    Also need heat pump Fujitsu AOU12RQ. I have inside unit.

  7. wjrobinson | | #7

    Useful blog Marc.
    Useful blog Marc.

  8. msamson | | #8

    Wish I was in an oil heat area
    Down from $8 per day to $2.25? I sure wish I was in an area that used oil. With the current low price of natural gas, even a COP 4 heat pump has about the same energy costs as a high efficiency furnace (which will become the standard next year)

    And that's even in BC- the only jurisdiction in North America with a carbon tax. I'd imagine a COP 3 ductless unit would fare worse.

    I'm also wondering, as an efficiency enthusiast, did you look at hydronic heat pumps, such as the Aermec and Dimplex units? I'd guess it would cost about the same to install as 2 ductless units.

  9. Expert Member
    Dana Dorsett | | #9

    Careful what you wish for...
    Wishing you were in a oil-heating environment? That's a bit odd- most of those folks are wishing they lived on the gas grid where the fuel is about 1/3 the cost per BTU.

    In western BC a pretty-good better-than-average Fujistu like Marc's will average a COP between 3.0-3.5. (See: ) But the only air-source hydronic solution that would touch that in a western BC climate would be the Daikin Altherma, and then only if a lot of upgrades were made to the radiation to get the water temps down to 100F or less (peak). The pre-existing fin-tube baseboard the oil-boiler was running simply wouldn't cut it. With the cost of the additional radiation required it would be well over 2x the cost of a 1-ton mini-split.

  10. Marc Rosenbaum | | #10

    @Eric - retrofitted basement wall insulation keeps the temp above 50F now
    @Curt - I figure DHW efficiency was barely over 10%. Everett Barber also figures DHW-only efficiency from boilers at 15-25% for more normal use rates.
    @Mark - DHW is separate. If I was using 0.6 gpd of oil for the 6 warmer months, that's about 100 gallons of oil, say at $400/year, and it doesn't count oil used during the heating season for DHW. We used under $180 worth of electricity in the heat pump water heater/Marathon tank in 12 months.
    @Mike - Dana answered the air/water HP question. More $, lower COP, and low service water temp, inadequate for existing fintube convectors.

  11. watercop | | #11

    Thanks - about what I figured
    Thanks - about what I figured

  12. dsmcn | | #12

    Consequence of distributed to single-point heat source
    You have moved from a distributed heat source (radiant heat in all rooms) to a single-point heat source (the minisplit in the living room). I am very curious to know the consequence in the more distant rooms.

    For example, if you were to leave the house for several days in "vacation mode" (thermostat at 50 degrees or so in winter), upon your return, how much longer would it now take for the whole house to come to an even and comfortable heat level? (And does it?)

    Similarly, if you simply close a bedroom door at night, how great is the variation in temperature by morning?

    It would also be helpful to see a floor plan for us to know how open it is.

  13. jwing | | #13

    Refrigerators are air
    Refrigerators are air conditioners, not heat pumps. Refrigerators do not have COP>1. Heat pumps are air condtioners that can operate backward. In using a compressor to extract heat from cold, outdoor air, waste heat is generated and that heat is expelled indoors.

    It is true that heat pumps add more energy to the space than goes through the electricity meter (when in the heating mode). It is also true that all forms of resistance space heaters are 100% efficient. But we need to go back to the source to find the whole effect on energy supplies and environmental degradation. Typical fossil fuel electrical generating stations are 35% or less efficient. In that light, a 85% boiler or furnace looks pretty darn good.

  14. GBA Editor
    Martin Holladay | | #14

    Response to J Wing
    J Wing,
    I disagree with your definition of "heat pump," and I side with Marc. The definition of a heat pump does not require reversibility.

    American English Dictionary: A heat pump is "a device for cooling or warming an enclosed space by removing heat from interior air and transferring it out, or by absorbing heat from outdoor air, or from a hot-water source, and transferring it in."

    Merriam-Webster Dictionary: A heat pump is "an apparatus for heating or cooling (as a building) by transferring heat by mechanical means from or to an external reservoir (as the ground, water, or outside air)."

  15. alyssjt | | #15

    hunky piping/sizing
    "The hunky looking white pipe in the photo below contains the refrigerant lines and wire that connect the indoor and outdoor units. (Normally this would be outside but the gutter interfered.)." I'm puzzled as to why people run the piping on the outside of the wall. Why not bring it in through the sill and into a more temperate basement area, then up through a closet or wall, then a short run on the interior wall over to the indoor unit. Wouldn't this provide an even greater degree of efficiency than having the piping on an outside wall? I have been heating a 24 x 40 Cape style log home with a centrally located woodstove for 25 years. My thought is that I can replace it with 2 ductless minisplits, one on each floor. I am concerned about sizing them properly. Is a blower door test in order, or can these just be sized via the typical volume/btu formula?

  16. GBA Editor
    Martin Holladay | | #16

    Response to Alice Twombly
    Q. "Why not bring [the refrigerant lines] in through the sill and into a more temperate basement area, then up through a closet or wall?"

    A. You could. In a retrofit situation, the decision has to be made on site. Routing the line depends on many factors, and every house is different.

    Q. "I am concerned about sizing [ductless minisplits] properly [for my existing house]. Is a blower door test in order, or can these just be sized via the typical volume/btu formula?"

    A. A blower door test is often a good idea, for a variety of reasons. But a blower door test is not required when you install a new heating or cooling system.

    The first step is to do a heating load calculation and a cooling load calculation. This is not the same thing as what you describe ("the typical volume/btu formula"). It is not based on just volume or square feet. It is based on many factors.

    The usual way to perform these calculations is with computer software. The best-known method is called the Manual J method. For more information, see these articles:

    Saving Energy With Manual J and Manual D

    How to Perform a Heat-Loss Calculation — Part 1

    How to Perform a Heat-Loss Calculation — Part 2

    Calculating Cooling Loads

    When Do I Need to Perform a Load Calculation?

Log in or create an account to post a comment.



Recent Questions and Replies

  • |
  • |
  • |
  • |