My house is heated by an old oil furnace, and the oil tank could spring a leak and flood our basement at any moment, so I'm hoping to replace it with a ground source heat pump this year. I'm trying to do my research now to figure out how big a heat pump I'll need, and how much energy it will consume.

First, there seem to be many different rules of thumb for estimating the size of heat pump required for a given space. I don't have a huge house, but the basement and garage are heated... EDIT: Here are some specific numbers:

  • One rule of thumb I read is (volume of heated space) x (indoor temp - outdoor temp) x .133. I figure our total volume is about 40,000 cubic feet (that's a 1,600 sqft footprint x about 25-ish feet of height, including the basement). We would want to set our thermostats to 68ºF in the winter, and the coldest days we expect to plan for here in the Boston suburbs are about -6ºF. So that works out to about 400,000 BTU.

  • Another rule of thumb: (total floor area) x (50 BTU/sqft for Massachusetts, in climate zone 5). Let's say our total floor area is 4,800 sqft (that is, 1,600 sqft x 3 stories, again including the basement). That works out to 240,000 BTU.

  • Using other calculators and various rules of thumb gives me a huge range between about 300,000 - 600,000 BTU (25 - 50 ton).

My gut instinct is that these estimates seem very high. Some of the comments already posted here before my edit suggest the same thing. I read this article, which suggests that I should not pay attention to rules of thumb at all, which seems like good advice. And Googling for typical residential geothermal heat pump sizes, it sounds like 3-6 tons is much more reasonable. So, my first question is: Why are the rules of thumb I initially used off by a factor of ~10?

Next, it's possible to estimate the power consumption of the system based on its published energy efficiency ratio (EER). Looking at the Energy Star data, it looks like it's reasonable to expect an EER between 16 - 19 BTU/watt-hour for a water-to-water, closed-loop heat pump. So, ignoring rules of thumb, supposing I have a 5 ton system producing heat at a constant rate of 60,000 BTU for every hour it's running (1 ton = 12,000 BTU, right?), it would be consuming about 3.2-3.8 kW of electricity. Let's meet in the middle and call that 3.5 kW. Note that my initial rule-of-thumb estimates had me consuming more like 30 kW, which would be an absolutely stupid amount of power.

Still, 3.5 kW seems a bit high. If I use that constantly all day, that comes to 84 kWh. For comparison, according to my electric bills, my house uses about 15 kWh per day total on average, maybe 30 kWh if I'm really blasting the air conditioning on a hot day. So I feel like I must be off by a factor of about 3-5. Or, to make this more tangible, my electric company charges about $0.28/kWh. So if I was blasting the heat constantly at 3.5 kW for a month, that would be about 2,500 kWh or which would come out to about $700 per month. That just has to be wrong; nobody would buy heat pumps if they were that expensive to operate.

I'm pretty sure the error here is that I won't be running the heat constantly at full blast; my thermostats will turn the system on and off as needed. But this adds an extra bit of complexity to the estimates that I don't know how to reason about; this is where I need help. My questions are:

  • How can I estimate how much time the system will actually be running? Can I estimate this somehow using my existing oil heating system?
  • Is it better (from the perspective of energy efficiency) to have a high-powered system that heats my space quickly and then shuts off, or to have a lower-powered system that heats my space more slowly and steadily?
  • EDIT: I just learned that variable-speed heat pumps are a thing. I had been assuming that heat pumps can only turn on and off like a furnace, but perhaps that's not true. I have no idea whether this is a common feature, or how thermostats work with a variable-speed heat pump.


  • 2
    You're confusing peak usage with constant usage. Just as your furnace doesn't burn 24/7, a heat pump doesn't run at maximum power 24/7. If you want realistic numbers, ask what's typical additional electric usage for your climate -- while remembering that individual choice of where to set thermostats and variation in house size and insulation also affects the numbers.
    – keshlam
    Jul 25, 2023 at 14:04
  • 6
    You don't mention the size of your house or the climate, but with the numbers I'm used to, 500k BTU is way over what you would need for a home. A 2000 sqft home is going to be in the 60k BTU (5 ton) range.
    – JPhi1618
    Jul 25, 2023 at 14:22
  • 2
    Why the heck did you neglect to share the square footage of your house? Your numbers seem absurd unless you're trying to condition a large restaurant or warehouse or something.
    – MonkeyZeus
    Jul 25, 2023 at 14:47
  • 2
    Key questions: Size of house (total square feet, how many stories), where are you (city/state should be enough - then we can look at various maps to determine an expected number of heating degree days or other factors), how old and, if very old, has it been upgraded in terms of double/triple pane windows, insulation, etc. With those can get a reasonable idea. But 500kbtu? Extremely unlikely. A different tactic: If you know how much oil you use in a typical year, that can give a rough idea (based on typical efficiency). Jul 25, 2023 at 15:02
  • 2
    And besides, 30kW even for part of the time would require 125A (or so) just for heating. Some people are crazy enough to do that just for water heating but I don't recommend it. Jul 25, 2023 at 15:04

3 Answers 3


To start with, please take an hour or two and review Technology Connections' series on heat pumps. Because it covers the fact that air-sourced heat pumps have gotten much better, and they no longer need "emergency heat strips" - and as a result, ground-sourced pumps are no longer necessary to get good performance in the snow belt. That could be a game changer costwise, but that will also affect performance.

You can't use "rules of thumb" to figure needed heat

Because they always grab for "easy-to-grab" data about your house, like square footage, which are not far from totally irrelevant. The number that matters is rather hard to obtain, because it relies on three things:

  • Insulation and sealing
  • Insulation and sealing
  • Insulation and sealing

These 3 factors (six?) dwarf other constraints. You could build a 300 square foot Home Depot shed in the back yard of a 4000sf modern LEED Passivhaus, and use more BTUs heating the shed than the house.

Your approach in your answer - looking at actual BTU expenditures using the fuel content of the oil and furnace efficiency. Good plan.

Now I am concerned with your presumptions about heat pump efficiency, because while they can hit 18 BTU/hr per watt, to do that air-sourced requires temperate conditions. Ground source can provide that if it's large enough.

Don't let the water heater hold you hostage

With hydronic systems, I often see people very, very reluctant to abandon hydronic because their system provides water heating as part of its function. I really dislike this setup, because it forces you to run the furnace all year simply for water heating, and that is not efficient at all. Consider switching to a unitary tanked water heater for under $600 - or better, a heat pump water heater - only about $1500.

The electrical load of a tanked water heater need not be large, even a plain water heater can have the heating elements changed to. 3800W, and heat pump water heaters are lower power still. The water heater can also be pre-empted so it suspends when another load is running e.g. the house's heat pump.

With water heat removed from the equation, you are free to make different decisions about whether to retain the hydronic system (which is incapable of A/C) or go with a split system. The head units are getting better and better, some hide in the ceiling and others use the form factor of a traditional radiator. All offer A/C, so that's inefficient window (or worse, portable) units that can go away.

  • Note that it's possible to have a hydronic system that can heat and cool, but it may mean having to scrap the existing boiler in favor of a reverse cycle chiller (i.e. a heat pump that uses water on the indoor side instead of air) as I'm not sure if a boiler will tolerate being exposed to chilled water during the summer. Jul 26, 2023 at 3:52
  • @ThreePhaseEel Yeah, I've wondered why you can't just run 35F water through the radiators and have a drip pan. Probably not enough of a temperature difference. My parents lived in a community where each apartment had a heat pump that interchanged with service water at ~70-100F. The service loop was gas heated in winter and cooling towers in summer. That could have as easily been ground sourced if enough ground water could've been obtained. Jul 26, 2023 at 4:01
  • Yeah -- radiators themselves don't work so well with 35degF water but fan coils are perfectly happy with such, and so are some other types of systems. Of course, then you have to deal with latent load control...(that said, if you have that down, you can do crazy stuff like radiant cooling of floors) Jul 26, 2023 at 4:02

A partial answer to my own question:

Following Solar Mike's suggestion, I went back and looked at my oil heating bills, and found that I burn an average of 4.8 gallons of #2 heating oil per day during the coldest winter months. This can be converted at a rate of 138,500 BTU/gallon, for a total of about 665,000 BTU/day.

My furnace is labeled (and, to my surprise, actually tested by the oil company) as being about 85% efficient. That gets the number down to about 560,000 BTU/day to actually heat the space. It's actually a bit less than that, since some of the heat goes to our indirect water heater, but I don't have a good estimate for how much.

Going back to the Energy Star EER data of 16 - 19 BTU/watt-hour, that equates to 29.5 - 35 kWh per day if I generated the necessary heat with a heat pump. That's much better than the 84 kWh I estimated above, and it's on par with the ~30 kWh I use for cooling in the summer, so I think I'm satisfied with this estimate.

Finally, a satisfying comparison: we paid about $23/day for oil last winter. If we convert to heat pump, we should expect to pay about $8.40/day for the electricity. And then we can start talking about solar panels...


You don't give the numbers you're deriving this from, so it is hard to say exactly which step you have messed up.

But for reference, I've been tracking total energy usage for my 1300 square foot house in ASHRE climate zone 5a, converting both electric kWH and gas therms to MBTU. I do have solar electric, which lowers my electricity usage by a variable amount and hasn't yet been added back into my usage numbers, but let's assume that's a wash across the time period for now.

Before putting in the minisplit and blowing insulation into the walls, long-term average energy purchase across multiple years was 74 MBTU per year.

Since those changes, and with tremendously more use of air conditioning, that number dropped to 51.25 MBTU.

Your mileage will certainly vary. And some of that difference could be year to year variation in weather, both in how hard the HVAC had to work and in how much solar was subtracting out. But this gives you one set of numbers to consider.

  • I've updated the post to include some concrete numbers; let me know what you think. Thanks! Jul 25, 2023 at 18:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.