The key question for this post is about as simple as it gets: If I have two choices for heat pumps, which one will use less electricity?
In my case, one option is the replacement ground-source heat pump that has been recommended, at a base installed price of about $25K per heat pump. The other option is to replace my dead ground-source heat pump with a modern air-source mini-split heat pump, at somewhere around half that cost (call it 60% after adjusting for likely difference in equipment life, in my particular case).
This is a stupidly hard question to answer well. As I explain at length below.
But, after doing all the homework that I care to do, for my house and my climate (with mild winters and an efficient gas-fired secondary heating system), the answer is that either style of heat pump (air-source or ground-source) will use roughly the same amount of electricity. Or near as I can tell, based on published data.
That’s not due to the underlying physics of the situation. If it were only about the physics, ground-source would win hands-down. Instead, that appears mainly due to faster technological improvement in air-source units over the past decade or so, compared to ground-source units. This seems to have fully offset the “natural” advantage of ground-source. In effect, my real-world choice is between air-source using the current generation of technology, and ground source using older technology. (The model of ground-source heat pump I have been offered was first introduced in 2016.) Or, at least, using a less-efficient design for the heat pump itself, disregarding which heat sink (air, ground) is used. That’s what makes it a tie ballgame, as of now.
This leads me to conclude that replacing one of my dead heat pumps with (e.g.) a name-brand air-source mini-split system:
- Is substantially cheaper, even accounting for likely shorter equipment life.
- Incurs no significant loss of efficiency compared to my ground-source option.
- As a bonus, bypasses my house’s barely-functional 1959-era ductwork.
Ground source systems still have some clear advantages. All the equipment is indoors, and so likely lasts longer. They work well even extremes of cold or hot weather.
But the fact is, there just ain’t that many of them, particularly in a relatively mild climate like Virginia. Of the roughly 4 million annual residential heat pump installations per year (in 2022), maybe 50,000 (call it 2.5%) were ground-source units. That has big implications for how rapidly the units reflect improved technology, and how much choice you have for who installs and services your unit.
Unless some unforeseen problem arises, I will replace one three-ton dead ground-source heat pump with a pair of 1.5-ton mini-split air-source heat pumps.
And I will not feel the least bit guilty about doing so.
I was going to give full and excruciating details but the overall accuracy of the conclusion does not warrant that. Below, I sketch out enough to summarize how I arrived at the numbers above.
SEER, EER, HSPF, COP, and all that jazz.
The efficiency of a heat pump varies, based on the how big a temperature difference it is trying to pump against, and how close you are to the maximum capacity of the system. The bigger the temperature difference, and the closer to maxed out, the less efficiently the heat pump runs.
This means that, despite what you read from many internet sources, you cannot simply convert one heat-pump efficiency measure to another with a simple conversion-of-units number. Yes, you must do that first, because some of these measures mix BTU/Hs and watts, and others don’t. But in addition, you also have to make some sort of adjustment for how stringent the test is.
It’s very much like EPA mileage. The MPG the EPA gets depends on how the car is driven. Typically, EPA city mileage is much worse than EPA highway mileage. If you compare the city MPG of one car to the highway MPG of another, you’re making a mistake. So it is, in spades, with SEER, EER, COP, and HSPF.
Now we get to the hard part: Things are hazy.
If you Google SEER, say, you’ll see the same zero-details definition everywhere: It’s the ratio of the cooling power produced (in BTU/H), to the electrical power supplied (in watts). But as to, how, exactly, that’s measured, it’s hard to find any information at all. E.g., is the energy used to run the water pumps included, what indoor and outdoor temperatures were used for the test, how were ducts, water pumps, etc. factored in, and so on.
- The details of the tests are proprietary and reside behind an expensive paywall.
- For the same measure, ground-source and air-source heat pumps use different methods.
- Certain aspects of overall energy use — duct system back pressure, water pump electricity use, and resistance electrical heating for backup heat — are either ill-specified, or not stated as to impact.
Among the things that I’ve seen hints for, but no definitive answer, is how these tests treat the waste heat of the electric motors themselves. I saw at least one credible-looking website showing that ground-source heat pumps add the value of this waste heat to their heating output, as if that heat would make it into your ductwork. But air source heat pumps do not. That’s consistent with where the compressor is located (inside for one, outside for the other). But it boils down to an assumption that the waste heat of the compressor motor somehow warms the air in your ductwork, which clearly isn’t the case for the units in my basement now. I have yet to find a clear answer on that, and it matters materially to the comparison.
So you need to take the table above with a grain of salt. My interpretation is that if there is a difference in efficiency across the three units I looked at, it’s small.
Definitions
Each of these measures compares output heating or cooling power, to input electrical power used.
EER (energy efficiency ratio). Cooling. Measured at a steady 35C outdoor air temperature, 26C indoor air temperature, and 50% relative humidity (for the outdoor air?). Heat/cool is measured in BTU/H, electricity is in watts. I think the test calls for the unit to run full-blast when this is measured.
SEER (seasonal energy efficiency ratio). Cooling. Near as I can tell, this is set up to simulate the range of temperatures you would see in a “standard summer”, so to speak. Heat/cooling power output is measured in BTU/H, electricity input is measured in watts.
COP (coefficient of performance): Heating: Generically, COP is simply watts of heat out, divided by watts of electricity used. Heat pumps have different COP values depending on the temperature tested, and how hard they were running. But the EPA-reported COP appears to be for one temperature, and I think its with the unit running full blast. Heat/cooling power is measured in watts, electrical input power is measured in watts.
HSPF (heating seasonal performance factor). Heating. Like SEER, this tests the units over a range of temperatures designed to be a sort of “standard winter”. I believe that, where the unit has a resistance-heating secondary heater, if that clicks on during the testing, the electricity used in secondary heating is counted toward the total. Heating power is measured in BTU/H, electrical use in watts.
The -2 suffixed versions of these appear to include a more realistic measure of the back-pressure of typical home ducts. Best I can tell, in the typical situation, you’d expect the (e.g.) SEER2 rating of an appliance to be 5% to 10% lower than the SEER rating.
Accounting for test stringency: SEER to EER conversion, units-adjusted HSPF to COP conversion. Here, I found some sketchy internet sources suggesting that where you have SEER and EER for the same unit, SEER is typically 85% of the EER value, due to the more stringent testing cycle. So I used that to adjust these all to a common EER-style basis.
Conclusion so far
Again, take this table with a grain of salt. There’s a whole lot I don’t know about the details of how each test is applied to each type of machine. And probably never will know, particularly for the details of testing ground source machines, where tests specifying outdoor air temperature are irrelevant.
That said, if you adjust for the difference in units-of-measurement (BTU/H versus watt), and assume that the tests that use a broad range of conditions (SEER, HSPF) tend to run about 85% of the equivalent tests that use a single set of conditions (EER, and COP as EPA reports it), then you get the comparison above.
Which, honestly, is just about what I came up with, back-of-the-envelope, when I first looked into this some years ago. The super-high-SEER Japanese-made heat pumps that emerged a decade ago seemed to eclipse (my estimate of) my existing ground-source heat pump’s efficiency. SEER 25? Maybe I mis-recall. But I do recall being startled with how high the available SEER ratings got, for air-source units.
Bottom line, efficiency-wise it’s a tossup. If I weight each units two numbers by local degree-day (3x heating a cooling), I get my estimated all-year efficiency values of 3.6, 3.5, and 4.0 for the three heat pumps examined, respectively.)
If your location experiences lot of time at extremely cold or hot temperatures, ground-source heat pumps still seem to offer some significant efficiency advantages over air-source. And, for sure, because the equipment is all inside, ground-source is likely to last longer.
But in my case — with a relatively mild climate, efficient (gas-fired) backup heat, and so on — it’s six of one, half a dozen of the other.
Finally, this pretty strongly suggests that the current tax law is out-of-date. The huge advantage given to ground-source heat pumps might have made sense in 2004. It appears to make no sense in 2024.
Once upon a time, ground-source heat pumps were king. But not any more. And the law has yet to catch up with that.
Winter approaches. 
