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

Why are COPs not higher ?

rhl_ | Posted in Mechanicals on

When it comes to heat pumps I regularly see heat pumps with COPs quoted at 45 degree outdoor temperatures in the 2.5-3.5 range. My question is what is the limiting factor in there designs that prevent us from having higher COP pumps?

I was inspired by the ideas in most recent post:

Where I live the ratio of gas/electricity prices for equal energy content is about 1:3. That is, 1 unit of energy in gas can be delivered to me for about 1/3rd the price of that same unit of energy as electricity.

However, interestingly, where I live, in the NYC area, the median half hourly air temperature on a heating day, is around 45 degrees (happy to share the analysis). So, I’m right on the cusp here, if a heat pump with a COP a decent amount above 3 the ratio of electricity/gas prices was to show up, converting away from natural gas would save me money on my energy bill.

Is is there any reason not to expect this soon?

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. _Stephen_ | | #1

    COPs are slowly creeping up. The theoretical maximums are a fair bit higher than what we see today, but manufacturing a reliable, inexpensive device capable of achieving them over a wide range of temperatures is challenging.

    There are however, minisplits that do fairly well however. The 9RLS3H Fujitsu minisplit has an HSPF of 14, which roughly translates to a COP of 3.8. However, that HSPF is for a warmer climate than yours. It'd be pretty close to 3 for you though I'd expect...

  2. rhl_ | | #2

    Right, but, what are the biggest engineering challenges? The NYC area is a _huge_ market, if the COPS were just a little bit better, then it would make *a lot* of sense.

    1. user-626934 | | #15

      Dana covered the engineering very well below...the technology to achieve >3COP heat pumps is already available and has been deployed in the U.S. (though slowly) over the last 10+ years.

      Switching from gas to electric heat pumps is not so much a technology’s a policy and “cultural inertia” issue...that is “people problems”. People problems are so much harder to fix than technological ones.

  3. user-2890856 | | #3

    I question your 1 / 3 ratio Ryan . Remember that 1 KWh is only 3,414 BTUh while NG is 100K BTUh . There are 29.29 KWh to equal a single therm of NG .

  4. Expert Member
    BILL WICHERS | | #4

    Natural gas is usually sold by the cubic foot, and natural gas has an energy content of just over 1,000 BTU per cubic foot. 1kwh of electricity is roughly equivalent to 3.5 cubic feet of natural gas by energy content.

    Note also that electric resistance heat is essentially 100% efficient — all input energy in kw is converted to heat, there is no loss. Natural gas heating isn’t 100% efficient, so if, for example, you have an 80% efficient furnace, you are using 20% more natural gas by energy content for the thermal energy you’re producing.

    I would say the big challenges are dealing with physics. I like to say “the one true law in the universe is ‘physics is a b****’.” Then on the engineering side, you have to apply physics at some cost point that people are willing to pay. This is not often easy to to do, especially as you start to approach theoretical limits.


    1. user-2890856 | | #18

      I disagree . Gas is usually sold by ccf or mcf or 100,000 or 1,ooo,ooo BTUh . This is also the number used for COP figures on equipment that uses such a rating

  5. rhl_ | | #5

    Yes, and it's about .25 cents a kwh or $7.30 for NG and its somewhere between $2 and $2.50 per therm. These numbers are estimates, arrived at by doing a least squares fit on my energy bills (summing supply and delivery), and then looking at the slopes. I did a fit over ~1.5 years of data.

    All of things aside if you take an electrically powered heat pump, and we divide it's COP by the ratio of electricity price/gas price, then if this new ratio is greater than the old ratio, we are in good shape.

    My boiler is *in theory* 95% efficient. For the sake of this argument i'm giving the boiler the benefit of the doubt. if it's less efficient (and can't be made more efficient), then switching is obviously a better idea earlier.

    1. AndyKosick | | #16

      Your motive for wanting to see higher COPs seems to be transitioning from NG. The other side of that equation is the price of NG. I would argue heat pump technology is already incredible (no reason to stop improving) and the price of NG is the biggest part of that problem; which makes it more of a political issue. The many long term environmental and health effects of fossil fuels are not accounted for in it's price. Not to mention the irreplaceable eXergy value of hydrocarbon fuels being squandered generating electricity and heating buildings to 70F.

      1. Expert Member
        BILL WICHERS | | #17

        Natural gas is not like oil. Many natural gas deposits have been shown to naturally replenish themselves over a relatively short period of time, by processes that are not well understood. Natural gas is not “irreplaceable”. I would argue the use of NG for heating and electrical generation is not squandered, but is rather and efficient use of a clean fuel. Be careful with any “data” regarding long term health effects too, much of it is politically driven or based on too many assumptions to allow for any valid conclusions to be made.

        Higher efficiency though is always a good thing. Regardless of the energy source, why waste any more than absolutely necassary?


  6. Expert Member
    Dana Dorsett | | #6

    COP is not a fixed number- it varies with temperature and is also a function of how much surface area there is on the coils for the air flow and refrigerant volume. For a given coil size there is an optimum range for heat transfer efficiency, and both the fan motor and compressor motor efficiency makes a huge difference. The seasonal AVERAGE efficiency of a 9RLS3 might be 3.8, but at 47F (one of the AHRI standard test temperatures for the HSPF testing) operated at an optimal low to mid-range speed it's COP can be 5 or higher.

    See Figure 5, page 10 (p18 in PDF pagination) of the bench tested COPs of an older 12RLS2 across temperature and at different compressor & fan speeds:

    Note that at minimum fan & compressor speed the COP is nearly 5 even at 35F, but half that at max speed.

    At colder temperatures the spread in efficiency with modulation shrinks, primarily due to the larger temperature difference between the outdoor coil and indoor coil. The vapor injection scroll compressors used in cold climate minisplits (and almost all Fujitsus) have made huge improvements in both capacity and efficiency at low temperatures compared to simpler compressors, but those compressor designs are usually optimized for peak efficiency at a particular temperature-difference range. I'm not aware of any US vendors of big-air handler heat pumps that are using this compressor technology, and the COPs and capacity below 20F drops pretty fast compared to any of the vapor-injection type compressor heat pumps.

    Using CO2 refrigerant has the potential for higher COPs at big delta-Ts, but are very high pressure and the refrigerant loops can't be tweaked in the field. Sanden's water heaters are using that technology, and can make 175F water with -20F air at a COP better than 1.

  7. rhl_ | | #7

    Hey Dana!

    Yes, your getting at the point i was alluding too in my first post, ill take a look at that article. Of course COP is a function of temperature, but, then it begs the question if we know the distribution of temperatures at the install site, can we estimate the distribution of COPS? This should be a straightforward weighted average. If for a fixed heat pump system (i.e. the coil size is fixed) COP(t) is say the worst case COP as a function of temperature for an hour of operation, then we can take \sum_{all temperatures t} COP(t)*{probability outdoor air is at temp t for an hour} and that should be a pretty decent estimate of average COP of the unit when installed. I went ahead and bought this weather data for myself. (Various sites will sell you about 10 years of history for about $5).

    Semi related -- can the sanden units be used for cooling too? I guess it makes sense that the refrigerants ability to condense would be the limiting factor. We really don't have better refrigerants?

    1. Expert Member
      MALCOLM TAYLOR | | #8


      "if we know the distribution of temperatures at the install site, can we estimate the distribution of COPS?"

      To generalize that a bit. You should be able to estimate the COPS based on geographic regions, or climate zones.

      1. rhl_ | | #9

        Do all climates zones have roughly the same temperature distributions ? The only useful generalization is the one where over a more broad region the results are all about the same.

        1. Expert Member
          MALCOLM TAYLOR | | #10

          No that's true. Even locally the temperatures experienced by a house here vary quite a bit based on the proximity to the ocean, or shading of the site. But generally it would be useful to be able to say that, for instance, on Southern Vancouver Island you can predict an annual COP of 4, whereas in the BC Interior you should plan around one of 2.5.

  8. Expert Member
    Dana Dorsett | | #11

    Different climate zones and location have different temperature distributions. Multiple organizations compile and publish hourly binned temperature data for use in heating system design. The 99 % outside design temperature is the 99th percentile temperature bin for the location, but tells you nothing about the distribution average.

    But more than just temperature, with modulating systems it's important to factor in the modulation level the heat pump would be operating across temperature since the COP with modulation level varies as well. It varies quite a bit at freezing temperatures and above. This is not a simple 2-dimensional problem- oversizing to hit a marginally higher COP at the 90% or 99% temperature bin may deliver a sub-optimally lower COP at the average wintertime temperature, and abyssmal average COP when cycling on/off a lot during the shoulder seasons. Running a COP of 5 at minimum modulation (as in the 12RLS2 example) gets destroyed if the compressor is spinning up from zero multiple times per hour, delivering an average of less than 3.

    As a general rule with modulating heat pumps it's best to size it at a capacity at the 99% outside design temp 0.9 x-1.5x of the load at that temp, but the total modulation range @ +47F and efficiency at min-modulation at +47F can push it one way or the other. For instance, a 3/4 ton LG can modulate down to about 1000 BTU/hr @ 47F, whereas the cold climate 3/4 ton Mitsubishi can modulate down to only 1600 BTU/hr. But the COP of the LG is pretty crummy at 1000 BTU/hr, well below it's optimum, and but the Mitsubishi's COP at 1600 BTU/hr is still excellent. Then it become a matter of figuring out the actual load at 47F and estimating where in the efficiency curve the heat pump would be operating, which ends up being a WAG in the end.

    The NEEP spreadsheet publishes the COP & capacity at both min and max output at +47F, +17F, +5F, and at the lowest rated operating temperature of the model for most heat pumps listed, which helps with the guesswork.

    The HSPF number is modeling it based on the average efficieny in ARI zone IV, (not to be conflated with DOE climate zone 4). The ARI zones are based on the number of hours per year at full load, which isn't always the best way to estimate the efficiency of a modulating system, but it's not a terrible way to make apples-t0-apples comparisons among models of similar capacity.

    1. fourforhome | | #13

      Where can we get the ARI zone IV variables?

      1. Expert Member
        Dana Dorsett | | #14

        The original specs were really based on fixed-speed or 2- speed simple compressors, and doesn't reflect the true as-used performance of reasonably sized (for their loads) fully modulating ductless systems using vapor injection compressors. In-situ monitoring of a fleet of cold-climate Mitsubishi FE12s in Idaho Falls (ARI zone V) done by the Northwest Energy Efficiency Alliance (NEEA) consortium 7-8 years ago showed the the fleet met or beat the zone IV HSPF tested numbers for that model. For fleet average to be that high, some individual installations had to have beat their zone IV HSPF by quite a bit, likely due to being more optimally sized to the actual loads.

        In situ testing by Northeast Energy Efficiency Partnerships (NEEP) on the same model (or maybe it was an FE18- same series but bigger) installed in an uninsulated enclosed porch/patio in southern New Hampshire a few years later (zone IV) came up well short of it's nameplate HSPF numbers, largely due to being woefully undersized for the load and running at it's maximum speed most of the season rather than modulating with load.

  9. user-723121 | | #12

    What are the options for heat pump backup in cold climates? Electric resistance heat, some type of thermal storage such as sub-slab? Whatever options there are, the building shell really needs to be top notch thermally and cold weather heat distribution is a factor to be considered. We could have all the thermal storage a basement could provide but there must be a system of distribution as the heat transfer through the floor is never going to equal the heat loss through the exterior walls and ceiling except maybe on the most very efficient homes. An open floor plan between floors may provide good mixing of air but only on very airtight, highly insulated structures.

    I have found very airtight, superinsulated building shells with adequate levels of foundation insulation including sub-slab have very uniform temperatures on all levels. Some of these have been open floor plans and I believe a slow and large convective loop may be at play.

Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |