Skip to content

Fighting Climate Change with Heat Pumps

Heat pumps are cool. When new refrigerant regulations take hold, they’ll be much cooler.

Today’s post is co-authored with Duncan Callaway.

Geez, it’s hot. We’ve been sweltering through record-high temperatures here in California. Our house is one of the 30% of California homes without air conditioning. And our furnace is getting older. If we replaced our furnace with a heat pump, we could get efficient heating and efficient cooling all in one. And we’d be doing our part to move the decarbonization-via-building-electrification ball forward. So we’re getting heat pump curious.

Joe Biden is excited about heat pumps. He recently invoked the Defense Production Act to ramp up domestic production, and the Inflation Reduction Act includes generous heat pump tax credits and rebates. Governor Gavin is also heat-pumped. He’s offering rebates to California households that will help him meet his target of 6 million new residential heat pumps by 2030. 

California is offering $3000 rebates for households switching from natural gas to a heat pump. Incentives under IRA are as high as $8000 for households below the median income.

Some recent research out of UC Davis finds that, for households that are installing AC for the first time, or households that need to replace their old air conditioner with a new unit, it makes climate sense to make the AC a heat pump and replace the furnace. We’re pretty convinced we have a heat pump in our future. For us, the question is not whether to heat pump…but when?

If we invest in a heat pump today, we could transition our home heating operations off of natural gas and onto California’s greener-by-the-year electricity grid. However,  today’s heat pumps use refrigerants that pack a serious climate punch. Thanks to some amazing efforts to address this problem, tomorrow’s refrigerants will be much more climate-friendly.

Energy nerds that we are, we spent a Saturday night working through this heat pump climate impact math. Though we’re convinced that the building electrification process needs to ramp up across the country, we’re now also convinced that we should wait a few years to join the revolution. These calculations also underscored, for us, the critical role of regulations that should soon shrink the carbon footprints of refrigerants used in heat pumps and air conditioners.  

*If you are curious about how we got the numbers we are throwing around in this blog post, go to the end for a wonky list of assumptions, calibrations, and equations.*

Refrigerant GWP is a BFD

Heat pumps use refrigerants to efficiently bring warmth into your house in the winter and remove heat from your house in the summer. Today’s residential heat pumps use R-410A which has a 100-year global warming potential (or GWP) of 2090. English translation: A ton of R-410A refrigerant traps as much heat in the atmosphere as 2090 tons of CO2. Yikes.

Heat pumps are not supposed to leak refrigerants. But they do. Leakage rate estimates vary. The UC Davis study we mentioned above assumed a 6% annual leakage rate and a 25% leak at the end of life when heat pumps are removed and refrigerants are recycled. This well-documented calculator from E3 assumes about the same annual leakage rate, but an 80% end-of-life leakage rate.

The heat pumps recommended for our house would use about 9 pounds of R-410A. Using the Davis study’s leakage rate assumptions, our hypothetical system would release just under 10 metric tons total CO2e over a 15-year heat pump life (see calculations for details). To put that into perspective,  the US EPA estimates that a typical gasoline-fueled car emits about 4.6 metric tons of CO2 per year.

As we’ll show in a moment, these heat pump refrigerant emissions can be more than offset over the life of the heat pump by not burning natural gas (and releasing the associated CO2). So if you need to install a new cooling appliance today,  you might as well install an efficient heat pump and get rid of your furnace at the same time. But if you can hold off for a few years, there are regulatory changes on the horizon that will make the climate math of heat pumps even better.  

Over the last decade, a determined coalition of environmental groups, regulatory agencies, and a key industry trade association has been tirelessly working to get super-GHG refrigerants out of air conditioners and heat pumps. They are succeeding. The United States has now committed to phasing down the use of these refrigerants. Here in California, by 2025, CARB will limit the GWP of refrigerants used in new residential air conditioners and heat pumps to 750 (or lower).

This graphic shows the relative GWP of different refrigerants that can be used in residential heat pumps. R-23 was phased out in 2020. Because R-32 efficiently conveys heat, manufacturers claim it can reduce electricity consumption as much as 10% compared to other refrigerants (an added benefit). Source: Daikin promotional materials.

We called up a few HVAC folks, and they’ve explained that heat pumps using a refrigerant known as R-32 will likely become the new low GWP standard. But these heat pumps are not on the market yet, and the building codes that will allow us to install one have not been updated yet. So we need to wait for the new rules and codes to kick in by 2025. And, in case you are wondering, we can’t install today and swap refrigerants tomorrow. We’ve learned that these new refrigerants require new heat pump technology.

If we conservatively assume that current leakage rates will not improve between now and 2025, we estimate that a heat pump purchased in 2025 will have a refrigerant carbon footprint of just over 2 metric tons over its 15-year life.  That’s a pretty big difference from the almost 10 metric tons that today’s heat pumps emit. 

So far we’ve considered the climate benefits of waiting. But there are also some climate costs of delay. Using the numbers for our home (see below), we estimate we’d be able to save about 1 metric ton of energy use-related emissions annually by switching from our natural gas furnace to a more efficient heat pump. These annual emissions savings would more than offset the leakage emissions from today’s heat pumps over a 15-year appliance life. But the climate benefits of waiting until 2025 easily exceed the climate costs of delaying. This has us inclined to postpone this home improvement project until we can get the greener goods.

Considerations, caveats, conclusions.

We are not suggesting that heat pumps are bad for the climate! If your AC craps out, and you need to replace it today, the climate will benefit if you choose a heat pump and replace your furnace. Here in California, targeting heat pumps at new construction and homes where folks need cooling now to adapt to intense heat today makes sense. 

We are suggesting that the refrigerant-related climate benefits of waiting could exceed the energy-related climate costs of delay for households who (like us) have some flexibility in terms of when they make their heat pump purchase. Of course, numbers will vary depending on the climate you live in, the size of your house, the fuel mix used to generate your electricity. See nerdy details below if you want to DIY your own heat pump climate impact calculations. 

Finally, after doing these calculations, we have a new appreciation for the high-stakes importance of phasing out high GWP refrigerants. EPA phase-out regulations have now been finalized. Further efforts are underway. California recently passed a law to support enhanced refrigerant recovery and reclamation (SB 1206). AB 209 invests $40 million to accelerate the adoption of ultra-low-GWP refrigerants. As we roll out incentives to get people heat pumped, these initiatives (and more!) will be needed to keep refrigerants on the job and not in the atmosphere.

Keep up with Energy Institute blog posts, research, and events on Twitter @energyathaas.

Suggested citation: Meredith Fowlie and Duncan Callaway. “Fighting Climate Change with Heat Pumps”. Energy Institute Blog,  UC Berkeley, September 6, 2022,


Assumptions, Equations, Calculations.  

Furnace energy use emissions:  We started by downloading our gas consumption data.  We averaged usage across summer months to estimate gas consumed for cooking (we have an electric water heater). Subtracting this from winter month consumption, we estimated an average of 218 therms of gas per year for space heating.  That works out to 1.3 metric tons of direct CO2 emissions per year, using PG&E’s published numbers for CO2 per therm. We then assumed a 3% leakage rate in the gas supply chain, a GWP of 30 for methane, and a higher heating value of 22,446 Btu/lb to estimate 0.4 metric tons of CO2 equivalent emissions due to leaks.  This totals to 1.7 CO2e metric tons of emissions due to operating our gas furnace for one year.

Heat pump energy use emissions:  Assuming our furnace is 80% efficient, and that whatever heat pump we might buy would have a coefficient of performance of 3.3, we would use 1,548 kWh of electricity per year to heat with a heat pump.  E3 estimates that the average marginal operating emissions intensity of California’s power system will average 0.39 metric tons of CO2 per MWh over the next three years. That works out to 0.6 metric tons of carbon per year of direct CO2 emissions for electricity use for our future heat pump.  To work out indirect emissions due to methane leaks to produce electricity, we assumed all California’s marginal electricity emissions are from natural gas.  Assuming gas plants are 50% efficient on average, 1,548 kWh of electricity would require about 105 therms of natural gas input to a generator.  Using the higher heating value, leakage rate, and methane GWP above, this works out to a CO2 equivalent of about 0.2 metric tons of emissions due to gas leakage from our electricity consumption.  This amounts to 0.8 CO2e metric tons of emissions due to operating a hypothetical heat pump for one year.

The punchline for energy-use-related emissions: Switching to a heat pump today would save about 0.9 metric tons per year in energy-related CO2 emissions over the next few years.  (If you don’t think gas leaks are marginal, the savings would be 0.7 metric tons per year.)

Refrigerant emissions: We had an HVAC company out to our house.  They recommended a system that delivers 36,000 BTU per hour (also known as a 3 “ton” system).  A number of sites (for example here and here) indicate that air conditioners and heat pumps should be charged at 2-4 pounds of refrigerant per ton.  We took the central estimate (3 pounds per ton) to estimate that our home’s heat pump would need 9 pounds of refrigerant.  

To estimate how much of that would leak over the life of the heat pump, we started with the same leakage rates in the Davis study, which works out to 112.5% of a refrigerant’s charge leaked over a heat pump’s 15 year life. (This clearly assumes the system will be recharged at least once in its life, since it leaks more than it can hold.)  For R-410a, with a GWP of 2090, that works out to 9.6 metric tons of CO2-equivalent emissions.  We found a few estimates (e.g. this one) that R-32 systems can be charged with 20-40% less refrigerant than R-410a.  So we assumed an R-32 heat pump 30% less charge than one with R-410a, which works out to 6.3 pounds for our hypothetical system.  Using the same percent leakage rate as for R-410a, and a GWP of 675 for R-32, this works out to 2.2 metric tons of CO2-equivalent emissions from an R-32 system.  

The punchline for refrigerant leak emissions: Switching to a heat pump today, versus waiting for the more climate-friendly heat pumps to come online, would increase refrigerant-related CO2e emissions by 7.4 metric tons of CO2-equivalent emissions over the life of the heat pump. 

32 thoughts on “Fighting Climate Change with Heat Pumps Leave a comment

  1. What would your calculations looks like if you installed a monobloc heat pump such as a SpacePak where all the refrigerant is contained within the external unit and water is used to transfer the heat/cold? I would think a factory built and tested refrigerant circuit would leak much less than a contractor built one (how much does your refrigerator leak?) You’d also have the ability to store energy to take advantage of off-peak savings. Take a look at Harvest Thermal or Stow Energy

  2. Great article. There’s one thing that’s not mentioned. Mini-split and variable refrigerant flow heat pumps where refrigerant piping is field installed are estimated to leak three to four times more on a per ton of cooling capacity basis than systems where the refrigerant is factory installed in a sealed unit. Unfortunately, these are the types of systems commonly used in residences. My conclusions are based on an excellent report by PAE Engineers for the City of Seattle, May 5, 2020 titled, “City of Seattle Refrigerant Emissions Analysis, GHG Emissions Calculation Methodologies”. This leads me to believe that better field practices could make a big difference.

  3. I really enjoy reading this blog. My two cents comment as a physicists, about heat and electricity; briefly that heat is less useful than electricity and therefore cheaper to store.

    Here is the argument:
    (1) there is an efficiency limit from thermodynamics about heat engines and their reverse (the heat pumps). Even so, a simple compressor is probably far from this limit. Besides refrigerants, other technology such as combined cycles, with multiple working fluids, or investments in heat exchangers or sources (e.g. ground source for buildings HVAC) may become cost effective eventually.
    (2) storing heat (or cold) is easy: you just need thermal mass (a water tank, a bag of sand will do) and a heat pump; because of thermodynamics (see (1)), reversing the cycle (with a heat engine) has an efficiency limit, so it is not a good electric battery, it may offer cheap opportunities to shift time of use.


  4. It’s a great analysis of a single residential heat pump. What about the CO2e of one big heat pump compared with the total CO2 of several small heat pumps? Which option is better? This analysis would be useful for commercial buildings, university campuses, residential communities, etc. Intuitively, centralized cooling systems (basically several buildings use one common big heat pump) could have fewer refrigerant emissions than distributed split units (each building has a small heat pump). However, if buildings’ cooling loads differ, saying that one building wants more cooling and another wants less, centralized cooling systems would use more than distributed cooling units. This kind of analysis would help transfer the concept of energy-efficient HVAC systems toward carbon-efficient HVAC systems.

  5. I have to point out a few areas where this can all go wrong. You hit the first one — the horrible emissions of R410A. From our field testing in over 100,000 heat pumps and air conditioners, the estimate of annual leakage is too high. End of life leakage is probably right.
    The second place is Heat pumps will build load in the winter in the evening and overnight where the Natural Gas, Coal, and “Imports” (read probably coal and NG) are 75% of the electrical generation.
    The third place is heat pumps will add load for houses that have no air conditioning now (30% by your numbers) This load will be biggest in when people will return home and set the thermostat to their desired cooling temperature — driving the heat pump to run continuously rather than cycle. This increase load takes place when the generation from natural gas, coal, and Imports is at a peak (63% yesterday 9/6/22)
    The fourth place is I have never before heard that the efficiency of a gas powerplant INCLUDING DISTRIBUTION AND TRANSMISSION being as high as 50%. As I recall it is something around 33%.

  6. “ If we invest in a heat pump today, we could transition our home heating operations off of natural gas and onto California’s greener-by-the-year electricity grid. ”

    That ‘greener’ electricity grid is already strained beyond capacity. Closing Diablo Canyon, removing hydro in the North state, regulatory barriers to new and old natural gas, plus the mandate for EVs sure won’t make the grid very accommodating to more heat pumps.

    • Electrifying heating, cooking and driving will not add to the peak loads–almost all of that load strain comes from already existing air conditioning. Electrifying will largely fill the troughs, using the excess solar generation, and reduce the average rate so we will get a double dividend on saving in energy bills.

      • Richard – replacing gas heating systems with heat pumps, especially where AC is not already present will add to both the summer and winter peak loads.

        • So we want to prevent those who can’t afford ACs from being relieved of heat stress? The fact is that the proportion of those new heat pumps that will be run as ACs will be small because those areas are largely outside of the climate zones needing AC. However, as the climate changes, those households would have installed ACs anyway, so the the net gain will be trivial. And again, the grid is not a static system and its evolving rapidly (and could do so even more rapidly if the status quo forces didn’t stand in the way.)

      • Richard – you make assumptions that are factually incorrect regarding peak loads. See John Proctors comments that specifically address peak loads, which conform with my experience.

        • Winter peaks are irrelevant where the summer peak is 50% higher as in California. While distribution networks may have to be upgraded in some places (and that can be offset with DERs), that has no impact on generation. And again, the incremental AC use in those 30% currently without AC will be small, except that climate change will have caused them to install AC in any case. (The climate isn’t static now nor mean reverting.) So the incremental demand will be negligible compared to the alternative, unless you believe that we should withhold AC from those households (often because they can’t afford it.)

          • How are winter peaks relevant in California (or the West) if you have excess capacity of nearly 50%, much of it hydropower? The issue is resource adequacy, not GHG reduction, that Roger was raising. We’re a looong way from needing more wintertime generation capacity (vs. more renewable energy).

  7. I will be installing heat pumps in my house,it was built in the late 70’s and is solely baseboard head,in northern michigan that is pricey in the winter,since there is no fucting and furnace i feel ir makes sense to go this direction.they will be ceiling cassettes in all rooms and will mist likely cost considerably less to operate in the wintet,added benifit is cooling in the summer, down side to the change to heatpumps are the are electric and will add to the grid use,positives are less fossil fuels being used for heating. It is a great alternative,maybe not for all situations but never the less a great alternative.

  8. I have become a big advocate for monobloc (self-contained) heat pumps in large part because of the refrigerant leakage reality you have quantified. I even cofounded a business to help them scale and make a big positive impact. Monobloc heat pumps:

    – eliminate field refrigerant connections, which cause the vast majority of leaks
    – improve safety for natural refrigerants like R290 by keeping flammable material sealed and outside
    – can provide heating, cooling, and hot water with one device, which reduces refrigerant volume and system cost
    – easily can incorporate thermal storage to shift thermal load to align with cleaner, lower cost power

    One still has to deal with refrigerant release at end of life and after major repairs, but these issues seem more tractable than dealing with quality issues at a huge number of field-installed connections.

    I actually have an R32 monobloc heat pump at my house in Berkeley that provides heating, cooling, and hot water for my home. It is fully permitted and keeping me cool on this very hot day.

    Does field routing refrigerant with flared fittings everywhere give anyone else pause?

  9. 3% methane leakage is too low. I don’t know how old that EDF paper is (there is no date one it!) but you just have to Google to find out that:

    “A new study has found that leaks of methane, the main ingredient in natural gas and itself a potent greenhouse gas, are twice as big as official tallies suggest in major cities along the U.S. eastern seaboard.”

    The EDF paper you cite uses EPA figure but: “It’s also much more than the amounts estimated by the Environmental Protection Agency (EPA). A 2016 report suggested methane emissions in the six major urban areas the researchers studied totaled only 370,000 tons. “It’s easy to say that the EPA inventory is low, but it’s not as easy to say why it’s low,” Kort says. One possible reason for this huge discrepancy: The EPA estimate includes leaks from the natural gas distribution system, but it doesn’t include leaks from homes and businesses. Those “beyond the meter” emissions could include, for example, tiny whooshes of incompletely burned methane from home appliances such as gas stoves, furnaces, and hot water heaters. Taken together over a city of millions, such emissions could be substantial.”

%d bloggers like this: