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
Building Science

Do Heat Pumps Work in Minnesota?

Yes! Gary Nelson does it without auxiliary heat

Do heat pumps work in Minnesota? (House shown is not Nelson's.) [Image credit: Energy Vanguard]

Yes, heat pumps work in Minnesota . . . and Alaska, Vermont, and Ontario, too. Today I’ll tell you about one particular house in Minneapolis, Minnesota, that’s been heated with a heat pump for the past four winters. It’s the home of Gary Nelson, founder of The Energy Conservatory, which makes the Minneapolis Blower Door and the Duct Blaster for testing air leakage in homes and duct systems.

First, fix the building enclosure

As you might expect from someone who has spent the past several decades quantifying infiltration rates, Nelson’s house is pretty darn airtight. He’s lived in the same house for a long time, and has done work on it more than once—most recently with a substantial renovation in 2017. They moved back into the house in November 2017, and have been using the heat pump ever since.

Here’s the current status of the enclosure:

  • Infiltration rate: 1 air change per hour at 50 Pascals (ACH50), 500 cubic feet per minute at 50 Pascals (cfm50)
  • Walls: R-30 to R-40
  • Ceiling: R-50
  • Floor: R-0 in the old part, R-20 with foam under the slab of the addition
  • Windows: Triple-pane, argon-filled, three low-e coatings

The result of improving the enclosure so much is that the heating load is very low. He didn’t do a formal load calculation, but he did know how much heat he was using before the renovation. His heating system then consisted of a Polaris water heater (fossil gas fired) and an air handler to distribute the heat. He found that the system ran almost continuously when the outdoor temperature was -10°F, the 99% design temperature, and the amount of heat he got from the water heater was 17,000 BTU/hr. Then he calculated that the load reduction from his enclosure improvements would be offset for the additional load from an addition that was part of the renovation, so he figured he needed a heat pump with a capacity of 18,000 BTU/hr.

Gary Nelson standing next to his 18k Fujitsu ducted mini-split heat pump
Gary Nelson standing next to his 18 kBTU/hr Fujitsu ducted minisplit heat pump, with tubes and wires for a lot of monitoring

So he put in a Fujitsu ducted minisplit heat pump with a capacity of 18,000 BTU/hr. And because he understands heat transfer and Minneapolis weather, he put this heat pump in with no auxiliary heat. Really!

Heating performance

Winter 2017-18: This was the first winter after the renovation, and the low in Minneapolis was -15°F. The system performed very well. Even though the outdoor temperature went 5°F below their design temperature, the 18k heat pump held the house at the 72°F setpoint.

Winter 2018-19: The outdoor temperature got down to -27°F. The house temperature got down to 62°F but they were away in Australia at the time. Nelson told me that if they had been home, they probably could have gotten the house up close to the setpoint with their body heat and by baking some cookies.

Of course, Nelson measures and logs everything, so he also knows how much heat the heat pump was pumping and how efficient it was. During that -27°F cold snap, he calculated that the heat pump capacity was 8597 BTU/hr (2.52 kilowatts) and the power consumption was 1834 watts (W). The coefficient of performance was 2.52 ÷ 1.834 = 1.37. For comparison, electric resistance heat has a coefficient of performance of 1.

When the temperature rose to -17°F, the heat pump output rose to 13,000 BTU/hr and the power consumed to 1959 W. The resulting coefficient of performance was nearly 2, or double what electric resistance would have provided. (And to think that some HVAC techs tell people to switch to emergency heat when the outdoor temperature drops into the 30s°F!)

Winter 2019-20: He had nothing remarkable to report. They didn’t have any weather cold enough to call for any kind of auxiliary heat.

Winter 2020-21: This was another winter that tested his decision to skip the auxiliary heat. Here’s what he wrote to me:

This February we had a few days in a row when it didn’t get above 0°F and was down around -17°F at night. There was very little sun, which I think is unusual when it’s this cold. I think it ran flat out for at least three or four days, and didn’t quite meet the setpoint. The third morning, I turned on the [electric] oven for an hour or so with the door open and then set it to 350°F with the door closed for much of the day. I’d guess we probably used 20 to 40 kWh of resistance heat.

Cooling performance

Nelson’s heat pump is sized just about perfectly for heating. In a place like Minneapolis, that means it’s oversized for cooling. Minneapolis does get humid, too, and the result is a house that can stay at the setpoint temperature easily but doesn’t get dehumidified enough. After two summers of dealing with muggy indoor air, he installed an Ultra-Aire dehumidifier in the summer of 2020. As a result, he “enjoyed much better humidity control” during the cooling season.

Yes, heat pumps can carry the load in Minnesota

Gary Nelson is a smart guy and knows how to calculate heat transfer. He understands the heating needs for his house. He can read and apply the specifications for the performance of a heat pump. And he’s proved in his home over the past four winters that heat pumps work just fine in cold climates. He sized his heat pump close to the heating load and even installed the system without any kind of backup heat. Yes, he needed a bit of supplemental heat for an unusual few days of cold, cloudy weather, but the 40 kWh of supplemental electric resistance heat he used might have added about $5 to his electric bill.

There’s no reason to be afraid of installing a heat pump in a cold climate if you’ve done your homework. You don’t need to be as conservative with the sizing as Nelson has been, and you can get auxiliary heat installed to cover those rare weather events that make it difficult for the heat pump to supply all the heat you need.

Going with a heat pump is a great idea, especially if you’re replacing a gas heating system. Unlike fossil gas, electricity is getting cleaner all the time.


Allison Bailes of Atlanta, Georgia, is a speaker, writer, building science consultant, and founder of Energy Vanguard. He is also the author of the Energy Vanguard Blog and is writing a book. You can follow him on Twitter at @EnergyVanguard.


  1. BirchwoodBill | | #1

    Good article, looking forward to the new book so I can place it next to the Sigenthaler books and the Manual J. Curious to know what is used for the DHW. The new air to water heat pumps can cool, heat and produce DHW. Water for heat only needs to be 83 to 103f, so DHW at 120F would require more BTU.

    1. GBA Editor
      Allison A. Bailes III, PhD | | #3

      Bill, Gary has a GE GeoSpring heat pump water heater, which you can see behind him in the photo above. GE has since stopped producing it.

      Thanks for the kind words!

  2. Jon_R | | #2

    > Yes, heat pumps work in Minnesota
    Some do, many shut off when it gets cold. With widespread adoption of backup resistance heat, the negative effect on the grid will make summer peak loads look trivial.

    > he also knows how much heat the heat pump was pumping

    How does he know? It's not like measuring the heat output accurately is easy.

    > -17°F, ...The resulting coefficient of performance was nearly 2

    There are various models, but looks to me like Fujitsu measured a COP of 2.0 at +5F (NEEP). So COP = 2 at -17F sounds wildly optimistic.

    1. GBA Editor
      Allison A. Bailes III, PhD | | #4

      Jon, Gary has his heat pump connected to a lot of wires and tubes to monitor temperature, pressure, and relative humidity. And he's got an Energy Conservatory TrueFlow permanently installed in the system. When you know the air flow rate, temperature change, and humidity change, calculating the amount of heat produced is straightforward.

      1. sprockkets | | #36

        My particular mini split was 1.3 at those temps. Now, that's a senville unit, so not as good as Fujitsu. However, when the temps dropped below zero, the system went from around 600cfm to around 400. I made the mistake assuming airflow was the same at below zero temps, and after correcting the mistake, cop went from 2 to 1.3.

    2. Expert Member
      Dana Dorsett | | #7

      >" ...COP = 2 at -17F sounds wildly optimistic."

      Agreed, but "...nearly 2... " is what it being alleged, not "... COP=3..." .

      The "...nearly..." part is subjective- a COP of 1.5 would qualify in some peoples' minds, 1.8 (still a credible number for best in class air source heat pumps) in others'.

      1. GBA Editor
        Allison A. Bailes III, PhD | | #8

        Jon & Dana, I should have put in the actual number from the beginning, I guess, but I did include the numbers you could use to calculate the COP at -17° F.

        Output = 13,000 BTU/hr x (1 kW / 3,412 BTU/hr) = 3.81 kW
        Input = 1,959 W = 1.959 kW
        COP = 3.81 / 1.959 = 1.94

        So, nearly 2 means very close to 2.

        Now, whether that's correct or not is another matter. Gary did tell me, "This seems pretty impressive.  I believe Manufacturer’s data only go down to -5 and my data agree pretty well with theirs at -5 and above."

        1. Jon_R | | #9

          Would be nice to see much more of the collected COP data. Including the exact model #, so we can find the manufacturer’s data.

          Can the few COP points that manufacturers list be accurately linearly extrapolated for lower temps? The data here suggests a suspiciously different COP degradation slope from 5F down to -17F vs -17F down to -27F.

          1. lance_p | | #42

            Don't forget, the heating capacity at 62F indoor temp is much greater than the capacity at 72F, as is the efficiency I'd imagine. Too bad no one wants to sit in a 62F house. :)

  3. user-723121 | | #5

    The new age heat pumps might be just what we need to highlight very efficient building envelopes. This really makes economic sense for homes off the natural gas grid for which there are many in MN. So much wind power is being added in MN, ND and SD electric rates may be stable for the foreseeable future.

  4. STEPHEN SHEEHY | | #6

    My 99% design temperature is a comparatively balmy 0°F. We've never had an issue over six winters with our Fujitsu heat pumps keeping up with cold weather. I'm surprised that the COP is as high as reported at very low temperatures, but it sounds like he's pretty organized about data collection. And even if some supplemental heat is needed once in a while, it's no big deal.
    Obviously, the key is a robust, efficient envelope because it allows you to cruise through the occasional cold snap without interior temperatures dropping much.

  5. _Stephen_ | | #10

    If he's running the oven to generate heat for the house, because his heating system ran for 4 days flat out without hitting setpoint, it really feels like he should have installed aux heat.

    1. creativedestruction | | #12

      The only downside to the electric oven backup is its manual. If they're not home, who cares if it can't reach setpoint as long as things don't freeze? The heat pump would have to completely fail for that to happen, and every other home in MN runs the same risk if a gas furnace fails when owners are away.

    2. Jon_R | | #13

      > he should have installed aux heat

      Or possibly a larger Quaternity heat pump, so that Winter load could be fully met and a dehumidifier might not be needed in Summer.

      Always do a proper Manual J/S.

  6. gusfhb | | #11

    >>After two summers of dealing with muggy indoor air, he installed an Ultra-Aire dehumidifier in the summer of 2020<<

    I know I am wildly oversimplifying, but a dehumidifier is basically an AC with a tiny fan. Having AC oversized for the conditions is not an unusual situation. I would think turning down the setpoint on the AC and then manually setting the fan speed very low[taking away its ability to reach the setpoint] ought to work.

    I say this many times in the questions, that people kind of expect the minisplits to just 'work' but they are just manufactured items, they do not understand the conditions around them. Using them for AC I tend to leave the setpoint and manage comfort with the fan speed. Sure it would be nice if the auto setting was perfect, but until they get 50:1 turndown ratio and wifi integration with the National Weather Service, we may have to deal with occasionally hitting the button on the remote.

    If I leave my 22k Mitsubishi to its own devices, I will sometimes wake up to a muggy living room, where it finally gives up and shuts the compressor off in the middle of the night. If I set the fan to low, I have to remember to turn it up in the morning or by mid afternoon it cannot keep up.

    1. Jon_R | | #14

      Agreed, the popular and wrong opinion is that over-sized AC provides inferior humidity control (as compared to right and under-sized heat pumps). But it's almost completely about CFM/ton. Get that right and even 2X over-sized inverter heat pumps work fine.

      1. frasca | | #16

        Jon R - just wanted to add that I frequently see your comments on threads related to AC equipment and dehumidification and I appreciate your perspective on the matter. The field research you have linked to (some outfit in Florida?) is helpful but beyond my ken, and you do a good job of simplifying it.

        I live in 4c now where I don’t have to think about dehumidification, but when my wife moves me back south I will likely be pinging you for advice!

        1. Expert Member
          MALCOLM TAYLOR | | #17


          I agree. I've learned a lot useful stuff from Jon about both humidity and pressure differentials.

      2. sprockkets | | #37

        That's because that rule applies to fixed capacity systems, not variable capacity.

      3. emma_vt | | #43

        Could you point me to more information about right sizing CFM/ton?

        I'm struggling with this exact question right now trying to size a ducted heat pump for the house we're building in Zone 6. Installer wants to go bigger so the heating is right sized, I'm inclined to go small so the AC is right sized and use the wood stove as needed. But if the larger unit will work just fine for cooling then that would be great.

    2. this_page_left_blank | | #15

      A dehumidifier is different in that it doesn't cool the space. It's not just about fan speed. If the load is 100% latent, then A/C alone probably can't manage the humidity. My heat pump is matched to both peak heating and cooling loads. I still can't keep the humidity below 60% a lot of the time, no matter what settings I use on the heat pump.

    3. sprockkets | | #38

      Can't recall if Mitsubishi units can measure humidity. Only one carrier midea unit I know of can.
      Realize the more your unit focuses on humidity removal, the more energy it will take and efficiencies nosedive. Also, you'll have more moisture inside and more chances for mold buildup. That's why they try to not do that. To remove humidity, the unit has to generate low refrigerant temps by high compressor speeds and run near 0 superheat.

  7. neilt | | #18

    Nice to see performance results from a cold area like Minnesota, but how would a similar system perform in 1 climate zone colder in Winnipeg, MB. Any insight?

    1. this_page_left_blank | | #19

      The design temperature in Winnipeg is about -33C. That's below any residential heat pump's operating range. However, remember that only 1% of the time in an average year will be at or below that temperature. A good, cold climate heat pump will still handle the bulk of the heating load. Supplementary heat would be a good idea to cover the times when it can't keep up.

      1. mark_rob | | #31

        Hi Trevor
        I'm currently building two small houses in Northern BC that are using Arctic heat pumps out of Winnipeg. They say that at -20c they have a COP of 1.7. In addition as a backup system for below -25, I'm using a storage tank with an electric boiler build in, again supplied by Arctic. The hydronic hot water is heating the house, preheating the domestic hot water and heating incoming kitchen exhaust make up air all from the storage tank.

    2. Randy_Williams | | #22

      I'm in Northern Minnesota, -30C/-22F design temp. We have a house that's been up an running on a Mitsubishi central ducted system (one of the PVA Multi-position air handler heat pumps, 24,000 BTU) with an integrated electric plenum heater for the past two winters. During this past winter's polar vortex, where temps didn't get above -18C/0F for nearly 10 days, the homeowner said temps never dropped more than 2 degrees below the setpoint in the home. They said there was a slight lag between the ASHP cutout temp and the electric plenum heater, but no major comfort issues.

      An issue we are having with the ASHP is a lot of condensation below the outdoor unit. Mitsubishi said this is not normal, but this is something that has happened both winters it's been operating, it is an unwanted condition we are going to have to address.

      1. user-723121 | | #23


        The house you are referencing is highly insulated and very airtight. Just as is Gary Nelson's house (from retrofit) in Minneapolis. To get good results from cold climate heat pumps in MN the building envelope will have to be high performance.

  8. PBP1 | | #20

    Thanks, great info and reminder "electricity is getting cleaner". Any chance of knowing the sq ft and cu ft and whether its a single level or multi-level house (to get an idea of exterior surface area to sq ft/cu ft, footprint, etc.) and attached garage or not? (I'm too lazy to back out cu ft from "500 cubic feet per minute at 50 Pascals" but maybe I'll give it a go, something like 1 ACH with 500 cu ft/min = 60 min/hr x 500 cu ft/min = 30,000 cu ft as 1 air change in one hour, hence 30,000 cu ft interior volume, assuming 10 ft ceilings yields 3,000 sq ft house?) 18,000 BTU/h for 3,000 sq ft in Minnesota?

    1. Jon_R | | #21

      > 500 cubic feet per minute at 50 Pascals

      AFAIK, a windy day (say 20 mph) can generate 50 pascals across roughly 1/2 of the house. So say 250 CFM and that it's a mild 0F outside. That's 19,440 btu/hr for the infiltration load alone. Then switch on a kitchen exhaust fan and double that...

      I'd want to see a lot more data from Nelson's house before forming any conclusions.

      1. PBP1 | | #28

        Hmm, so have the blower door test on a calm day ;-) Or, have multiple blower door tests under multiple wind conditions to see impact of wind on things like sliders, maybe an argument for the aerodynamic passive house, on a spindle to always point into the wind.

  9. daveh4 | | #24

    Interesting that both this case and the cases discussed in the comments all used ducted systems. I would expected ductless would be the preferred (and better performing) system here. Any thoughts? I'm considering this for Hudson Valley (East Coast - Zone 4/5) and was working on the assumption that ductless was the only way to achieve subzero temp performance.

    1. Expert Member
      MALCOLM TAYLOR | | #25


      I don't know much about this. Can you flesh-out why not having ducts would affect the cold weather performance?

    2. Jon_R | | #29

      Everything I've seen indicates that ducted mini-splits and ductless mini-splits are similar in performance. Just a little more energy to move air around.

    3. sprockkets | | #39

      You lose energy in the ductwork, but that may not matter if it is in the living space. Otherwise your system loses its punch the longer the line sets are.

  10. daveh4 | | #26

    I'm definitely not an expert here, which is why I phrased it as a question. But besides the inherent losses you'll find in even the best ducted systems, the HSPF and SEER ratings of mini-splits are much higher than central (ducted) heat pumps. It's pretty hard to find central heat pumps with HSPF of 10, which is required to be a certified Cold Climate Heat Pump. The HVAC contractors I talked with all said this as well (don't always trust what they say though). Anyway, I would love to hear if anyone has researched information on this.

    1. Expert Member
      MALCOLM TAYLOR | | #27

      It's an interesting question. Luckily just an academic one for me as our climate here is so temperate.

    2. charlie_sullivan | | #30

      If you go to the NEEP database:!/product_list/

      you can find plenty--literally thousands--of centrally ducted models with HSPF > 10, some > 12.

  11. Nola_Sweats | | #32

    [also posted in Q&A]: If Minnesota doesn’t need auxiliary electric heat, do I even need to install heat strips in New Orleans? This is less about the $75-ish cost of the heating filament and more about how quickly even a 4.5kw heat strip would drain a battery backup system. NB: running the electric oven seems like cheating.

    1. charlie_sullivan | | #33

      Same answer as in Q&A: For you, the only reason to have them is that if you have a mechanical failure of your heat pump--in the compressor or perhaps a refrigerant leak--the heat strips could, at high electric cost, keep the house warm while you wait for service. But they can also kick in automatically by mistake, using lots of energy without you knowing, if there's a control failure. I would either install them and leave the circuit breaker off, requiring a manual action on your part to activate them in an emergency, or omit them.

      1. Expert Member
        Dana Dorsett | | #35

        >"For you, the only reason to have them is that if you have a mechanical failure of your heat pump--in the compressor or perhaps a refrigerant leak--the heat strips could, at high electric cost, keep the house warm while you wait for service."

        The other reason for strip heat is for comfort during defrost cycles. It's a common feature in ducted heat pumps to engage strip heat during defrost to avoid blowing tepid/cold air out the registers.

        But 4.5 kw (~15,000 BTU/hr) of strip heat is enough to cover the actual design heat loads of most code or better-than code houses in LA. Installing a switch between the heat pumps strip control and the heat strip itself to be able to enable it if/when the heat pump fails isn't insane, and not a huge cost adder.

        Have the heating and cooling loads of this house in New Orleans been carefully calculated with a Manual-J? Most AC and heat pumps that get installed are way oversized for the actual loads, and installing a right-sized modulating heat pump could be the best of all worlds, even without strip heat during defrost.

        1. Nola_Sweats | | #41

          Charlie: I have heat strips on the upstairs (bedroom) unit, so I won't freeze if I'm without heat downstairs while I wait for a compressor repair.

          Dana: I hadn't thought of the use of electric heat strips in defrost cycles! Near as I can tell, those are just cold-weather a/c cycles with the compressor fan off, and it does blow cold air inside if the strips are turned off, and I keep the current ones off. I think I'll get the strips to avoid the cold-air wedgie during the defrost cycle. I can always lock them out if we switch to battery backup ... if I'm awake.

          Manual J will be done once I've settled on a contractor. They ballpark a tonnage for the free estimates, but the two finalists both say they will do the full calculations. They know I know.

  12. mangler66 | | #34

    That COP is impressive, I don't get much lower than -17F here, so i would get about 2.0 or more. Does that include the blower? If so, is that a distinct air handler and what are the specs?

    The one thing i was looking for through the entire article was the model number for the Fujitsu. Did I miss it? I can't be the only one wondering.

    1. sprockkets | | #40

      I'd take that cop with a grain of salt. Assume it is 1.37. I say that because actual air flow may be reduced at those temps. Also, at those temps the compressor has to work really hard.

  13. ErikOlson | | #44

    Wow, impressive! May I ask why Fujitsu over Mitsubishi ducted heat pump?

    Am planning a highly insulated post frame house (1600 ft^2) in 55312.

  14. Kj_Feinberg | | #45

    Temperature extremes inhibit a mini split system's ability to function effectively. Extremely high and low outdoor temperatures make it difficult for an air conditioner to eject or absorb heat, resulting in a decline in both performance and efficiency.

    Typically, this decline is not perceptible during hot summer months, but dramatic temperature drops can drastically affect a mini split's ability to heat and cool during winter months.

Log in or create an account to post a comment.



Recent Questions and Replies

  • |
  • |
  • |
  • |