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
Guest Blogs

The Carbon Footprint of Minisplits

Is heating with a minisplit heat pump really ‘greener’ than heating with fossil fuels?

Which method of space heating has the largest carbon footprint? The answer depends on the carbon intensity of the power grid.

It’s often presumed that heating with high-efficiency heat pumps has a lower carbon footprint than heating with other equipment (and often it is). But how do you really know?

Do the math!

How much CO2 is emitted by my state’s electric utilities?

Since the EPA has promulgated its Clean Power Plan (CPP), Americans can, at least for the time being, make reasonable estimates of just how much carbon is emitted by their electricity use.

But first, some basics on the carbon emissions of different fuels. The U.S. Department of Energy’s Energy Information Agency has tabulated a short list showing the pounds of CO2 per million BTU (MMBTU) of source fuel. The most common space heating fuels on that list are:

    Heating oil:         161.3 lbs/MMBTU

    Propane:             139.0 lbs/MMBTU

    Natural gas:         117.0 lbs/MMBTU

Heating appliance efficiency matters

To reach a reasonable first estimate for an apples-to-apples comparison, it’s important to factor in the heating equipment’s efficiency. Better class oil-burning appliances have an AFUE (efficiency rating) of about 87%, while better-than-average appliances the burn propane or natural gas burners are in the 95%+ range. So, after adjusting for appliance efficiency, the carbon emission numbers look like this:

    Heating oil:     161.3 / 0.87 = 185.4 lbs/MMBTU

    Propane:         139.0 / 0.95 = 146.3 lbs/MMBTU

    Natural gas:     117.0 / 0.95 = 123.2 lbs/MMBTU

To determine the heat pump numbers for comparison purposes, start with the heat pump’s HSPF specification (an efficiency rating). An HSPF number can be converted into an efficiency percentage by dividing the HSPF by 3.412 (the number of Btu in one watt-hour of electricity). For example, a heat pump with an HSPF of 8.0 has an efficiency of 235%.

If a heat pump is properly sized for the heating load, the HSPF rating will be reasonably close the actual in-use efficiency for Climate Zone 4. In Zone 3 and warmer zones, the in-use efficiency is likely to be a bit higher than the HSPF, while in Zone 5 and colder zones, the in-use efficiency is likely to be a bit lower than the HSPF. You can assume a 0.85 multiplier to determine the degradation of a heat pump’s efficiency for each climate zone colder than Zone 4. (There are exceptions to this rule of thumb, but a close examination of the topic is beyond the scope of this article.)

For example, if the HSPF of a heat pump is 11.3, the “adjusted” efficiency of the heat pump when installed in Climate Zone 6 (two zones colder than Zone 4) will be something on the order of:

    0.85 x 0.85 x HSPF 11.3 = 8.2

The units of HSPF is BTUs per watt-hour, but electricity is metered and billed in increments of kilowatt-hours (=1000 watt-hours). So an HSPF of 8.2 means on seasonal average basis it delivers 8,200 BTU per kWh of electricity used. From there we can calculate the kWh input per MMBTU delivered:

    1,000,000 / 8200 = 122 kWh/MMBTU

Utilities in each state have carbon targets

Meanwhile, the CPP has defined carbon targets for every state (except Vermont, which was exempted due to a dearth of fossil fired power generation in that state). The carbon targets include the 2012 estimated carbon intensity per megawatt hour (Mwh = 1,000 kWh), as well as a projection for 2020 in a “business as usual” without the CPP, and a target for 2030 with the CPP.

So, let’s compare how well a heat pump fares in say, the Climate Zone 6 portion of Michigan, a state with a grid with medium carbon intensity:

    2020 Projections (without CPP):     1,588 lbs/MWh = 1.588 lbs/kWwh

   1.588 lbs/kWh x 122 kWh/MMBTU = 193.7 lbs/MMBTU

Under these circumstances, the heat pump is has a slightly higher carbon footprint than an 87% efficient heating appliance that burns #2 heating oil. (The oil-burning appliance has a carbon intensity of 185.4 lbs/MMBTU of heating.) When you include the electricity used by the air handlers, pumps, and burners, it’s probably a wash. But the carbon footprint of the heat pump is considerably higher than the 123.2 lbs/MMBTU of a gas-burning appliance.

But assuming that Michigan hits its 2030 target under the CPP (and remember, by 2030, a heat pump installed in 2016 would have reached more than half its anticipated lifecycle), the heat pump will be at 1,169 lbs/MWh, or 1.169 lbs/kWh.

At that point the heat pump will have a carbon footprint of:

    1.169lbs / kWh x 122 kWh / MMBTU = 142.6 lbs/MMBTU

That’s considerably better than an appliance burning #2 oil, and right in there with the 146.3lbs/MMBTU for condensing propane, but still well above that of natural gas.

But for comparison, take a look at parts of Maine that are in Climate Zone 6:

    2020 Projections (without CPP):     736 lbs/MWh = 0.736 lbs/kWh

    0.736 lbs/kWh x 122 kWh/MMBTU = 89.8 lbs/MMBTU

Heat pumps make sense in “low carb” states

Right away it’s obvious that in this state with a lower-carbon grid, the heat pump is “lower carb,” far and away, than any appliance burning one of the common fossil heating fuels. And the heat pump will be even lower carbon going forward.

Clearly, in states like Wyoming with a high-carbon grid, even when the grid reaches the 2030 CPP target, a heat pump won’t beat a condensing appliance that burns natural gas. But that’s not to say it’s impossible to get there in other ways. In deregulated electricity states like Michigan, it’s possible to buy grid-supplied electricity from greener sources. A 100% renewables package purchased through brokers or a direct power-purchase agreement would provide electricity for the heat pump that is essentially zero-carbon power.

Wyoming has very favorable wind resources — resources that could be developed if there is a retail market for 100% wind power — but state regulators in Wyoming promote vertically integrated utilities, and haven’t yet decoupled electricity markets. Wyoming also lacks net-metering for behind-the-meter (residential) PV, so until the regulatory framework changes, most heat pumps in Wyoming are going to have a higher carbon footprint than condensing gas, even if the state meets its CPP targets.

This isn’t intended to be a precise model of what’s going on; it’s just a reasonable rough cut. There are many factors that can raise or lower a heating system’s efficiency, and often local utilities within a state have much higher or lower carbon output per MWh than the statewide averages. But as a quick estimate, it’s way better than a guess.

Dana Dorsett has lifelong interests in energy policy, building science, and home efficiency. He is currently an electrical engineer in Massachusetts.


  1. AlanB4 | | #1

    Is the upstream carbon
    Is the upstream carbon included in the carbon dioxide calculations, as i understand it natural gas has some production and distribution carbon outputs

  2. DavidJones | | #2

    Properly Sized?
    I recall several GBA articles that suggested oversizing Mini -splits actually increased seasonal efficiency because the variable speed unit operated in its more efficient zone more of the time. The listed COP was an average, and it varied according to the load. I recognize that oversizing is generally considered a bad thing in any other type of equipment. Was the question of increased efficiency for Mini splits ever resolved either way?

  3. Expert Member
    Dana Dorsett | | #3

    Upstream carbon clearly not included.
    The EIA's short list of lbs-carbon per source-fuel MMBTU has none of those subtleties, just the pure chemistry. The refining #2 oil has a significant greenhouse gas footprint too, which varies by the feedstocks used.

    The upstream carbon footprint of fossil fuels for fired power is not included in the EPA numbers either.

    There are many ways one could attempt to refine the model, but it's not clear that those refinements would reduce rather than increase the error. This approach is simply a first order rough estimation. The as-used AFUE and as-use HSPF of the equipment will also vary from the tested performance by quite a bit, as does the power used by the air handlers, pumps, & controls of the combustion equipment, and the distribution losses from ducts & plumbing can vary wildly with the system design & execution (eg: leaky ducts installed in vented attics above the insulation.)

    Then there's the issue of whether your state will actually meet or beat the CPP targets. It's likely that many or most states will beat their 2030 CPP targets on the raw economics of how rapidly wind and solar power costs are falling. The CPP targets were published before the tax credit on wind & solar were extended out to year 2022, which will in the short term drive a more rapid build-out of PV and wind, but in the post tax credit scenario PV and wind will be by far the cheapest form of new power. The 2015 levelized cost analysis by the investment bank Lazard indicates that even before the tax subsidy utility scale solar is already at parity with combined cycle natural gas, and wind is already cheaper than all fossil fuels:

    In the new extended tax credit scenario I would expect most states would hit their 2030 goal by 2025 or sooner, and it wouldn't stop there.

    When the rough cut numbers on carbon footprint are close, use your best judgment. When they're not very close at all the direction is pretty clear. What is clear is that site-sourced power with PV sufficient to cover the power used by the heat pump makes it net carbon free. But there is more impact to install that PV on a high-carb grid such as West Virginia than it is in low-carb heavy-hydro Washington State:

  4. DavidJones | | #4

    Mini split makes plus renewables makes sense
    The article does a great job of explaining the carbon foot print of mini splits without renewable energy and it does mention grid level greener sources of electricity. One of the biggest advantages of Mini splits is that they are easily coupled with renewable sources of electricity. Mini splits can be easily paired with solar PV at the house to get a zero carbon foot print. Getting to zero isn't possible with the cleanest fossil fuel burned in the most efficient appliance.

  5. user-1041981 | | #5

    Mini-Split on PV
    Just a tangential thought from one EE to another. While one may live in an area with a high-carbon fuel based electric power generation, one can put PV on or next-to their house (except maybe in some states like Wyoming) and power their mini-split with carbon-free (ignoring the manufacturing of the panels and inverters) generated electricity.

    I'm sure you thought of that - just wasn't clear to the casual reader.

  6. Expert Member
    Dana Dorsett | | #6

    oversizing minisplits is sometimes a problem (@ David Jones )
    While upsizing by as much as 50% can sometimes improve efficiency on mini-splits, that's true only if it's not so oversized that it spends most of it's time cycling on/off rather than modulating with load. The modulation range is not infinite, and some models have much bigger turn down ranges than others.

    It's a good idea to run the heat load calculations at both the 99% outside design temperature, but also at +47F, which is a defined temperature in the HSPF at which both a minimum and maximum capacity is measured. If the minimum capacity @ +47F spelled out in the submittal sheet is dramatically higher than the calculated load @+47F the mini-split may be too oversized to hit the efficiency numbers.

    For instance, the 3/4 ton Mitsubishi can modulate down to 1600 BTU/hr @ +47F, with an 11:1 turn down ratio, whereas the somewhat higher capacity Fujitsu 3/4 tonner only modulates down to 3100 BTU/hr @ +47F, with a 7:1 turn-down:

    Even thought the HSPF numbers are on the Fujitsu are higher than the Mitsubishi, if your heat load at +47F is 2000 BTU/hr you'll actually do better with the Mitsubishi, since the Fujitsu would be cycling on/off whenever it's above freezing outdoors. But if your heat load at +47F is 4000 BTU/hr, you'll probably do better with the Fujitsu.

  7. GBA Editor
    Martin Holladay | | #7

    Response to David Jones and C.B. (Comments #4 and #5)
    David Jones and C.B.,
    In some parts of the country, you might try to run your ductless minisplit from a PV system. That works better in the summer (for air conditioning) than in the winter (for space heating), unless you live in Arizona.

    In cold, cloudy, northern states, the electricity needed to run your minisplit during the winter (for space heating) will be coming -- mostly -- from the grid.

    The sun doesn't shine when you need the space heating, unfortunately.

  8. DavidJones | | #8

    PV and Mini split
    In December my company just finished a 1400 sq/ft HERs -12 house in Connecticut. We have Net Metering in CT which allows us to generate electricity all year to offset the winter months. (Fuzzy math I realize since the grid power is generated by fossil fuels).

    While we did have a warm winter this year, we found the house performance exceeded the modeling. The 7.6 KW PV system made enough power to heat the house most days. We only had 2 weeks in the entire winter where we had a power deficit (looking at it on a weekly basis). We were still using the grid on cloudy/snowy days and at night.

    So we did use the grid powered by fossil fuel, but we more than offset our use with PV power and now the house is accumulating a sizable power credit. From a carbon foot print stand point, i don't think burning any fossil fuel in any appliance could possibly compete.

    In the past I had always advocated for solar hot water first, high efficiency gas fired appliances to pick up where solar hot water left off and solar PV as a last resort. Today with the cost of PV much lower and the simplicity of an air source heat pump, I think the all electric house makes sense.

  9. GBA Editor
    Martin Holladay | | #9

    Response to David Jones
    I'm a big fan of grid-connected PV systems and net metering. I'm delighted to hear about your success.

    The fact is, however, that a house with a good thermal envelope usually doesn't need space heat when the sun is shining. Under sunny conditions, the solar heat gain through the south windows usually keeps the heating system from coming on.

    The heating system comes on in the early evening, and also runs on cloudy days (when there is no solar heat gain through the windows).

    There are exceptions to this basic rule, of course. On very cold sunny days in winter, you'll often need heat in the early morning hours. But the sun doesn't really rise high enough in the sky to begin generating useful electricity until 9:00 am in winter.

    Here's what I'm saying: in northern climates, the need for space heat and PV generation are non-simultaneous. So the ductless minisplit is getting its electricity, mostly, from the grid.

  10. Expert Member
    Dana Dorsett | | #10

    The whole "run the mini-split off PV" thing.
    This can sort of work in locations with modest heat loads and decent wintertime sun, but only with high mass construction, and you'd better have a wood stove or something as backup.

    I recently walked & talked an urban-off-gridder in Texas on how to optimize the house design using thermal mass so than he could somewhat over-heat the place during the day in winters and still be comfortable on a 25F early morning, using a mini-split as the primary heat, with a small wood stove as the Hail Mary resource. I couldn't talk him into a rammed earth or adobe solution- he's leaning toward poured concrete with exterior foam, with a conventional looking brick veneer that fits in with the look of his city neighborhood. (There is both a natural gas & power grid in his neighborhood, but he is adamant about not hookup up to the grid rather than getting jerked around by changing utility rate structures regarding net metering. He'll be on city water & sewer though.)

    A high-mass construction approach might even be a reasonable approach in eastern Wyoming or Nebraska, but it becomes more difficult in cloudier, snowier or much colder places. But net-metered works almost anywhere.

  11. DavidJones | | #11

    mostly from the grid

    Agreed that on a minute by minute basis, the electricity is coming mostly from the grid.

    However all the grid power is being off set by PV power, which while not perfect seems better than simply buying from the grid and not off setting it.

    The article concludes that a mini split can reduce carbon foot print (if the grid power is clean enough) as compared to the best natural gas appliance. Taking that one step further by off setting the grid power entirely with PV power seems significantly better than a mini split without PV, from a carbon stand point. Unless I am missing something?

    I have considered using the Tesla storage system in conjunction with the PV and the min split. That could eliminate all dependence on the grid, but I think that the carbon benefit would be small because the PV power back to the grid is replacing fossil fuel energy.

  12. user-1119494 | | #12

    this would be a nice idea for a web-based calculator
    At least for folks who have the interest, but not enough interest (or confidence) to do their own calcs...

  13. Expert Member
    Dana Dorsett | | #13

    The carbon economics of grid power vs. PV
    The best thing you could do for your planet in a high-carb grid environment would be to stay grid attached, since every kwh you're putting onto the grid is taking that much off the high-carb inputs. Staying off-grid when the grid is available uses more resources (such as batteries) that have their own environmental hit and usually results in power curtailment to some degree from your own PV array (more power than you need, not enough battery to pack it away.) PV-to-battery-to-inverter is a lossier process than PV-to-inverter-to-grid too.

    Behind the meter batteries for grid tied systems make sense if the grid operator has control over the battery and can use it to provide ancillary services to the grid such as frequency control or line voltage stabilization, but don't make financial sense for the PV owner unless they are being assessed demand charges based on peak power draws. Demand charges common feature in commercial rate structures, but extremely rare for residential (though I believe there is a utility in Arizona nicking residiential PV owners for demand charges.) In markets like Australia where the residential rates are high, and there is no retail net-metering there is sometimes a financial rationale for behind the meter batteries, and that's probably the largest market for the Tesla (and a half dozen other battery manufacturers.) In CT, where the grid reliability is high and the net metering is at retail it'll never pencil out- buying a 2kw Honda as backup for the hurricane events makes more sense.

    Under New York's regulatory revisions currently under way there may soon be structures in place where aggregated behind the meter resources can be bid into the ancillary services, capacity, and localized marginal price wholesale markets, but that day isn't here (yet).

  14. GBA Editor
    Martin Holladay | | #14

    Off-grid vs. grid-connected
    Dana, you're right, of course. As someone who has lived off-grid for 41 years, I agree completely. Grid-connected PV is better for the planet (and a homeowner's pocketbook) than off-grid PV.

  15. qcoder | | #15

    PV and minisplit
    We also have PV, a nameplate 10 KW system, and minisplits. And just installed a GE Geospring 50 gallon heat pump electric ("hybrid") water heater, which we intend to use on heat pump mode. Check it out:

  16. GBA Editor
    Martin Holladay | | #16

    Response to Robert Opaluch
    You are correct that the AFUE rating does not account for the energy used by a heating distribution system (energy used for fans or pumps, or energy losses in the distribution system).

    However, the AFUE rating does account for energy losses associated with venting (chimney losses) and energy losses through the furnace jacket or boiler jacket.

  17. Robert Opaluch | | #17

    Overstated efficiency of burning fossil fuels for home heating
    Thanks Dana for your many contributions to GBA and the high performance building profession.

    A fossil fuel furnace or boiler doesn't heat a home. It heats air or liquid. A heat distribution system is used to move that heating energy through the building. I see minimally insulated pipes with insulation hanging off, located along poorly insulated (or uninsulated) basement exterior walls. Although these pipes could be called radiators themselves, there are heating system radiators to consider, too. Radiators are located along poorly insulated exterior walls, often under windows. The alternative (often leaky, uninsulated) air ducts, move heated air inefficiently, too. What's the efficiency of moving heat from a furnace or boiler into the habitable rooms of a home? How much goes into basements, crawl spaces, and out through exterior walls? What's the convection losses of heating under windows, losses through those windows to the outdoors, before heating the rooms? Granted, too tough to measure or even estimate precisely. But we could design a realistic, cost-effective high efficiency distribution system to estimate losses to add to the furnace or boiler efficiency number. The point being: Furnace burning efficiency is a vast overestimate of the heating efficiency of heating homes with fossil fuels.

    I think its fair to say that the public is being conned by these fossil-fuel heating efficiency numbers. They were intended to compare among furnace alternatives, not to estimate home heating SYSTEM efficiency. We need to consider the efficiency of the entire home heating system, not just the furnace or boiler efficiency. This doesn't seem to be included in calculations of fossil-fuel system efficiency vs. minisplits.

    Ductless minisplits dump warmed (or cooled) air right into a central point in the living space. Simple, direct. No complicated heat distribution through basements, crawl spaces, etc.

    Another related point: You must vent those poisonous fossil-fuel burning by-products. There are heat losses from fossil-fuel system flues, masonry chimneys especially. These losses also are not included in fossil fuel efficiency calculations, but should be, its a REQUIRED part of a fossil fuel-burning system. And not needed with minisplits.

    After adding in some reasonable estimate of the the heat distribution system losses and fossil-fuel flue heat losses, we could have a truer "apples to apples" comparison of the efficiency of a fossil fuel heating SYSTEM vs. minisplits. Could you add some rough estimates to make these adjustments (assuming they are not included already). I don't have any quantitative estimates myself.

    BTW I'm not a fan of minisplits (no pun intended), but IMHO they are far better than fossil fuel home heating systems for many reasons.

  18. Robert Opaluch | | #18

    Fossil fuel venting losses
    Martin, thanks for the prompt update!

    So I assume that the AFUE rating includes the losses due to heating air that goes out the flue, but doesn't include conductive losses through the structure of the flue (and related construction) itself, or convective losses along those poorly sealed masonry chimneys. Heat from the building is lost through the flue in addition to the air exhausted. Could be minor but with a masonry chimney may be more. Its extra baggage of a fossil fuel (and wood-burning) heating system. Doesn't apply to minisplits (or other electric) heating that produce no poisonous gases as a byproduct.

    There are so many nails in the coffin of fossil fuels, except they refuse to die! :-)

  19. Reid Baldwin | | #19

    Interaction of PV and heating type
    In terms of the carbon impact, I don't see the choice to install PV and the choice between fossil fuel heating and heat pump heating as interrelated. Changing from fossil fuel to heat pump will increase the load on the electric utility and the electric utility will respond by burning more of whatever fuel it uses. Suppose that without PV installed, the heat pump increases my demand from 5 to 10. If I have PV putting out 15 units, my demand would increase from -10 to -5. (I am using made up numbers with no particular units as an illustration.) My heat pump increases the utility's carbon emissions by 5 x whatever their mix is, independent of whether or not I have installed PV. My PV system decreases the utility's carbon emissions by 15 x whatever their mix is, independent of whether or not I have a heat pump.

    There may be some regulatory policy interactions. For example, the amount of PV I would be allowed to install under a net-metering program would be impacted by my choice of heating fuel.

  20. DavidJones | | #20

    Heat pump is related to PV choice

    Under most net metering regulations the goal is to size the PV system to cover but not exceed demand. You would not have a situation where the PV was producing 15 units while your demand was 5 or 10 units.
    If you burn fossil fuels for heating, you produce carbon. If you use a mini split for heat and produce the required amount of electrical power with a PV system, you have not added any additional carbon to the atmosphere.

  21. Dana1 | | #21

    Response to #16 & #19
    Robert: As stated in the blog piece-

    "There are many factors that can raise or lower a heating system’s efficiency..."

    AFUE testing is a reasonable approximation of net system efficiency when:

    A: The system is no more than 1.7x oversized for the 99% design load, and

    B: The heating system (including heat distribution) is fully inside of conditioned space.

    These are conditions often unmet in-situ, and yes the amount of electric energy use is not included (though comparatively small as a fraction in all but the worst system designs.) But as long as those two conditions are met the rest is somewhat "in the statistical noise" of measurement error. Flue losses (independent of flue type) are registered 100% as losses, even though there may be some heat transfer recovered from the flue to conditioned space depending on how the flue is routed ( exterior vs. interior.) Convective idling losses of flues have for that past 40 years or so been mitigated to less than 1% of the total source fuel energy by designing in automatic flue dampers, etc, and those are indeed accounted for in AFUE testing.

    With most fossil burners the AFUE test numbers come pretty close to the equipment's steady-state combustion efficiency, and as long as you're not short-cycling the thing to death even 3x oversizing doesn't move the needle on hot air equipment. With hydronic boilers efficiency losses due to oversizing are a bit more pronounced than hot air, less so with modulating boilers.

    So, the efficiency numbers are not a con, it's a rough apples-to-apples comparison when reasonable assumptions are met. But it's easy to screw that up.

    HSPF numbers on modulating heat pumps are actually much squishier than AFUE numbers, and more sensitive to oversizing issues than hot air fossil burners ( as discussed in #6.) The as-used efficiency is also sensitive to placement of both the indoor and outdoor coils, and more sensitive than fossil burners to clogged filters. Improper charging levels can also create large variance with the stated efficiency. The more idiot proof they make them, the more creative the idiots become. Mini-splits may be somewhat less screw-up prone than some fossil burner solutions, but they're far from immune.
    Reid: The grid mix is not static over the course of a day. During the heating season the overnight demand in most cold-climate areas is low, and the power used is from higher efficiency slow ramping baseload generators or must-run-constantly nukes. During solar output hours in the summer there is often high air conditioning demand being met by lower efficiency fast ramping oil or gas peakers. In general the PV output is offsetting a higher carb daytime mix, and the mini-split in heating mode is using a mostly lower carb mix. If you're putting up enough PV to cover the total power use of your heat pump, you are trading nighttime carbon use against similar or higher daytime carbon reductions.

    But the mix changes day to day, week to week, and year to year and place to place. In the cooling dominated southern US the absolute peak grid loads are at night, due to the prevalence of electric resistance heating in those region. A mini-split may be somewhat higher carbon in heating mode and the PV isn't offsetting as much, but it's still offsetting a higher carb daytime grid mix on average.

    In New England within the ISO-NE grid region more than half the overnight grid supply on a given day is from zero carb sources, but averaged over the year natural gas alone is over half the total grid mix, but mostly from combined cycle gas oprating at ~50% thermal efficiency. If you took the same gas and burned in a condensing boiler it would have 95% efficiency. But taking the 50% efficiency power out of the combined cycle generator and turned it into heat with a heat pump with a COP of 2.5, it would have 125% net thermal efficiency. (This is part of why when you run the crude carbon calc outlined in the blog bit it comes out advantage- mini-split in all New England states.) But if you're offsetting that New England natural gas used at the power plant during daylight hours with PV, your net carbon foot print will be negative at least for now.

    As PV becomes more prevalent the daytime mix will become mostly PV, so it will be offsetting less carbon 15 years into the future than it does today. The CA-ISO "duck" curve or HECO "Nessie curve illustrates the point.

    If & when YOUR state is producing more PV power on an average wintertime mid-day than the total grid load, installing PV will no longer be offsetting your heating power use. But when that day comes it will be both carbon & cash advantageous to be charging an electric car with the mid-day surplus (sometimes at negative wholesale electricity pricing) rather than commuting on fossil energy. For most states that day is more than a decade away.

  22. JIM BAERG | | #22

    The atmospheric contribution of Natural Gas
    As homeowners, we don't usually get to choose which fuel to use. The choice usually is between electricity (PV or utility supplied) and some form of gas or liquid fossil fuel. The carbon footprint of electricity depends on what mix of coal, Natural Gas or hydro is used to generate the electricity. Of the fossil fuels, Natural Gas has been successfully marketed as the most "green" option. Coal is clearly nasty stuff.
    Natural gas is mostly methane and methane, though more short lived, is 18 to 54 times as damaging to the atmosphere as carbon. Methane comes out of gas and oil wells and, after cleaning, is piped directly into the pipeline system. Thousands of pipe fittings are involved. The amount of leakage has not been systematically studied, but the spot testing has, in some cases, shown large amounts of leakage. It is entirely possible that Natural Gas is just as dangerous as coal when used for generating electricity and heating our homes.
    In the mean time, insulate, upgrade and install PV.

  23. Reid Baldwin | | #23

    Response to #21
    Suppose that every kwhr used to run a heat pump produces X carbon and that every kwhr produced by a rooftop PV system reduces carbon production by Y. Dana's argument in #21 is that X is not equal to Y. That does not dispute my point. My point is that X is not dependent on whether or not I have a PV system on my roof and Y is not dependent on whether or not I have a heat pump.

  24. Reid Baldwin | | #24

    Incremental mix
    One simplifying assumption in the analysis methodology of the article is using the utility's average generation mix. What really matters is the incremental mix - what the utility will use more of if additional load is added. If this distinction is small, the simplification is ok. If this distinction is large, then the analysis is over-simplified.

    Trying to figure out the incremental mix is much more complicated because it varies by time of day and probably a whole bunch of other factors. I doubt if hard numbers are available for most utilities. One can look at incremental mix in either the short term or the long term. In the short term, the question is: with the utility's present generation assets, what would they use more of to respond to additional load? Most types of renewables are not dispatchable, so they cannot use more of those. If the load is predictable, they may run the base load plants a little harder. Otherwise, they will respond with peaker plants. In the longer term, the question is: how would additional load influence the utility's decisions about building new (or retiring old) generation assets? If the additional load is mostly at night, they won't respond with more solar.

  25. Expert Member
    Dana Dorsett | | #25

    The EPA is not ignoring methane leaks (@ Jim Baerg)
    The amount of methane release from the oil & gas industry IS being methodically studied, and EPA regulations around that leakage is tightening.!OpenDocument

    But the prescriptive is still valid: "In the mean time, insulate, upgrade and install PV."

    From a cost-effectiveness point of view, efficiency (of all types, including the thermal performance of buildings) is still broadly cheaper than PV. Small scale PV is still more expensive than most other forms of wholesale electricity, but is rapidly crossing retail cost boundaries. The next few years will sort out whether/when net metering at retail loses viability in the face of high market penetration of distributed solar.

  26. DavidJones | | #26

    Response to 23

    With net metering, you can't realistically put power into the grid except to offset power that you have used from the grid. The heat pump justifies / allows you to use the PV. No heat pump means no electrical power consumption which means no capacity to send power back to the grid. Net metering doesn't allow you use PV to offset a unit of energy used by someone else. It only allows you to compensate the grid for the unit you use.

  27. Expert Member
    Dana Dorsett | | #27

    Of course it's not a high precision model! (@ Reid)
    There is never going to be sufficient data to estimate the exact overlap of grid energy use minute-by-minute that's going into a particular heat pump at a particular location within the state. The simplifying assumption is there to make a rough judgement of where it will be. In high-wind but still high carbon states the incremental power is sometimes wind (particularly in Xcel's area, where they've mastered the art of using wind rather than gas peakers to track load) sometimes gas (usually combined cycle, not peaker) or something else.

    Combined cycle gas plants are easily used to manage increased overnight load averages due to increased prevalence of heat pumps on the grid, and that makes it more akin to the grid average mix, not less. The increased peaker plant use argument is inherently untrue. Overnight peaker use is typically for frequency regulation and sometimes for dealing with short term unanticipated variance with load projections. Grid operators factor in the weather (and aggregate heating loads) into the projections, which is then bid into the day-ahead markets. Unless there is a severe capacity shortfall it's unusual for peakers to make it in the day-ahead market. In the heating dominated states the peak grid loads that sometimes approach generation capacity limits are the summertime peak air conditioning loads, not wintertime peak heating loads.

    Peaker plants are eventually going away- sooner now that the Supremes upheld FERC Order 745, which enables aggregated demand response to be bid into wholesale electricity markets. Between demand response and the introduction of even modest amounts of grid storage the capacity factors of peakers has been steadily shrinking. David Crane (former CEO of the large merchant power generation company NRG) recently predicted that the last peaker plant to ever be built in the US will be put into service prior to the year 2020, and that projection was made prior to the FERC Order 745 decision. It's cheaper (and more responsive) to provide the same grid services as a peaker with a combination of demand response & grid storage.

    Without the simplifying assumption the estimates become impossible. As I stated, "This isn’t intended to be a precise model of what’s going on; it’s just a reasonable rough cut." The actual carbon footprint of the power use won't be anything like twice, or half what the rough estimate comes up with, unless of course the CPP targets get beat resoundingly (which can in fact economically happen in some or most states.) In the 15-25 year service life of furnace or boiler things can (and will) change, and one can re-assess that later.

    Offsetting power use with rooftop PV does indeed offset the carbon footprint of the combined system, even if there isn't 1-1 precision in carbon due to the temporal displacement between the load and the power generated. It's roughly the same as buying renewable power through a power retailer to offset the carbon use. And true, it's independent of whether it's a heat pump or a gas-burner- one could also offset the carbon footprint of the gas burner with PV or a renewables PPA, but it would be a much bigger array, even in high-carb states, and would require remarkably high levels of electricity use.

    BTW: NREL recently published an analysis showing that something like 40% of the nation's power could be sourced by distributed rooftop PV:

    This isn't a prediction of what will happen, only the technical feasibility. But as PV costs relentlessly continue to fall, it may very well come to pass, once regulatory obstacles are cleared. The time scale may be fairly rapitd relative to the CPP timeline targets, which would of course render the whole crude analysis moot.

    If anybody has a better crystal ball than mine, can I have a peek?

  28. Reid Baldwin | | #28

    As is often the case in forums like this, there is much more argument than there is disagreement.

    David, if you read the last paragraph of my post #19, you will see that I agree that using a heat pump increases the amount of PV I am allowed to install. My point is that this is driven by regulations and utility policy rather than physics. From a physics standpoint, my PV system doesn't care whether I have a heat pump.

    Dana, I did not intend to imply that using average generation mix instead of incremental mix makes the analysis not useful. I was trying to get a sense for how much accuracy we lose with that assumption. I understand that we need to make simplifying assumptions to make analysis feasible. Before we make decisions based on the analysis, we should critically evaluate the assumptions.

    I am trying to use this analysis to help me make a decision. I have settled on a traditional gas furnace for heating. My next decision is whether to use a pure air conditioner or specify a heat pump at the capacity needed for air conditioning (which is much less than my design heat load). The heat pump would allow me the flexibility to heat with electricity during part of the year. I already concluded that the heat pump would never be efficient enough to save me money on my energy bill at current gas and electric prices. The next question is whether it would reduce my carbon footprint. This analysis says probably not, at least without adding PV into the analysis.

  29. Expert Member
    Dana Dorsett | | #29

    Accuracy at that level wasn't there in the first place.
    If anything the overnight carbon footprint of the heat pump is slightly lower than the daytime use in most markets. At current PV deployment levels the difference between day & night carbon footprint of the grid isn't usually huge, but it will evolve as more PV goes onto the grid.

    The ISO-NE grid operator posts the grid mix online, updated every 5 minutes along with the daily load projections, localized marginal price(s), etc here:

    Right now it's the mid-day lull, down in part from the AM and PM peaks due to behind-the-meter PV not being metered and sold at the LMP, with about 40% of the total coming from natural gas. Both the hydro & NG fractions of the pie change with the demand/load, but the NG fraction does most of the swing, almost ALL of which combined cycle. The snapshot right now could easily look the same at midnight as it does right now. The renewables fraction moves around with the wind, with maybe 1% coming from fully metered grid-scale solar on a sunny day (click on the renewables tab to see the break down) , but year on year that fraction is increasing. But in the late afternoon of a so-so July day fully 2/3 of the power cookie is usually gas, but at midnight the same day the share can drop to half or less. In winter the AM grid peak load occurs well after the heat load peak you'd see with a modulating heat pump, and after the PV is starting to deliver.

    You'll see that on a sunny day like today in New England there's a bit of the CAISO duck or HECO Nessie thing going on today if you look at the system load graph on that page. Only some of that is from behind-the-meter PV, but in a decade the mid-day dip will be much more pronounced on sunny days with low heating & cooling loads like today.

    But that's the full regional mix- you'd have to consult with your local utility to know for sure where THEIR power is coming from.

    It's usually worth the up-charge for heat pumps rather than straight AC. Even if it isn't big enough cover your space heating load, it can often supply 100% of the shoulder season load, usually at higher efficiency than it's HSPF numbers indicate (unless your shoulder season includes +17F outdoor temps or cooler). Plenty of folks New England with oil fired space heating bought mini-splits to offset $4/gallon oil, and manage to stay warm most of the heating season on mini-splits alone even when the capacity only covers half the load at the 99% outside design temp.

  30. charlie_sullivan | | #30

    Incremental mix and time of use
    It turns out that ISO-NE publishes incremental carbon emissions data, and even breaks it down by day vs. night, though not hour by hour. 2014 is the most recent available.

    It's actually worse at night than during the day: 931 lb/MWh day and 949 night (10 pm to 8 AM). The average emissions/kWh is almost surely better at night, largely because of a heavy reliance on nuclear, but the marginal is never nuclear day or night, because that's run at full power whenever possible. You can see on p. 15 that the marginal generation is usually gas (72%), which explains the similar numbers day and night, but it's sometimes coal (8%), sometimes oil (4%) and sometimes hydro or pumped hydro. My guess is that the marginal is more often coal at night than during the day, and more often hydro during the day than during the night, and that explains the slightly higher marginal emissions at night. But that difference is small enough that it's not worth worrying about--the main story is both are dominated by gas and both are about the same.

    More relevant to the heat source decision in New England is the difference in marginal rate in the winter vs. the summer. Fig. 5-7, p. 25, also below, shows that the rate is 1200 in the winter, vs. 825 or so in the summer. On p. 13, that reason for that is explained: there's more coal and oil used in the winter, because natural gas prices are higher in the winter, because of pipeline congestion. Also, there's a little less hydro available in the winter because of lower rainfall.

    With the 1200 winter marginal rate, for New England, Dana's calculation lands at 146 lbs/MMBTU, right on par with propane, worse than direct use of natural gas and better than direct use of heating oil. Lacking state-by-state marginal data, that makes a heat pump a good choice where oil would be the alternative, and probably a good choice vs. propane as well, because we can expect a lower carbon grid in the future.

    Deciding between natural gas and a heat pump in New England gets a little funny, because if you choose natural gas, you are participating in causing pipeline congestion that is what causes the grid operators to move to coal and oil. So just looking at the two independently is less defensible. However, once you start talking about adding PV, that's clearly beneficial, because the pipeline congestion time scale is long enough that the daily timing of the PV contribution doesn't really matter. By adding PV, you are reducing winter electric energy consumption, and thus helping move ISO-NE towards being able to meet the electric demand within the pipeline capacity without resorting to coal and oil.

  31. Dana1 | | #31

    The rate of evolution matters
    Data from the 2014 analysis is already out of date, since there has been closings of both nukes and coal fired generation in the ISO-NE region, and an increase in both wind and combined cycle natural gas. Even the 2020 business as usual projection under the CPP will likely turn out to be an exaggeration. Now that FERC Order 745 is now settled, the impact of demand response on wintertime gas grid constraints will also become obvious well before 2020.

    Viewing the wintertime natural gas peak load gas supply constraint issue as a pipeline capacity problem would be a mistake, since that thinking drives policy toward a more expensive and financially risky solution- more pipelines. Site storage is a cheaper and more targeted way to manage peak gas-grid loads driven by the concurrence of higher power use from heat pumps competing with higher gas use from space heating equipment. But note, even best gas burners out there will be delivering thermal efficiency in the mid 90s percentage range for space heating, whereas a combined cycle plant might hit the low 50s. But leveraged by a heat pump at a fairly low COP of 2 that 50% net thermal efficiency becomes 100% for the mini-split, a number not achievable condensing gas.

    In the Massachusetts case the current governor, Charlie Baker (R), and his energy secretary, Matthew Beaton, have been pushing for an imported hydro solution to the state's energy grid congestion and carbon footprint issues, purchasing hydro power from both Quebec and New Brunswick (both of which usually have plenty of surplus) in lieu of building out more large scale gas grid. That can happen both more cheaply and more quickly than building out pipelines, and would have a competitive marginal load price, and would lower the wintertime spot prices & availability of natural gas to the local generators. It's not a done-deal, and there are competing interests with different visions of the energy future, but this is one of several solutions that would bring MA below the 2030 CPP targets well before 2025. The offshore wind proposal by DONG (Dansk Olie og Naturgas, the largest offshore wind developer in the world) and the not-really-dead-yet Cape Wind project could also put a huge dent in the overnight carbon footprint in that time frame,

    Bottom line, the future isn't written yet- legislative decision making still matters,and focusing too closely on the carbon footprint of the marginal load may quickly be moot as automated demand response becomes the go-to cheapest resource for relieving grid constraints (gas & electric).

    As an aside, MA Energy Secretary Matt Beaton built one of the first certified PassiveHouses in Massachusetts, heated by a 2.5 ton multi-split:

  32. charlie_sullivan | | #32

    Agreed--marginal is not the whole story
    Agreed--the marginal analysis is not the whole story. And as Reid pointed out, there is short term (e.g. hour by hour) marginal analysis, which is what the ISO-NE report discusses, and there's long term marginal analysis--the decisions about what gets built. And there's also the prediction of the future short term marginal analysis. I suspect that even if the average carbon analysis has changed relative to 2014, the marginal may not have. But as you point out, the future automated demand response alternative changes the picture. (I'm not quite sure how to evaluate the marginal carbon impact of adding a load when curtailing other loads is the immediate impact--I guess it's then the marginal generator's emissions in the hour that that load gets shifted to, which gets a little trickier to trace.)

    Here in NH there are people who are not thrilled about having an additional transmission line run through this state on the way to MA, but I hope they work out a compromise. Demand response can take care of a lot of the variability of renewables, but it would be a lot easier with some new transmission as well. I think environmentalists should be promoting new transmission instead of opposing it.

  33. Dana1 | | #33

    2014 is history, but not ANCIENT history yet.
    The gas supply constraints that drove wholesale spot prices for both gas & electricity stratospheric during the cold snap of February 2015 won't happen again, or at least not in the same way. The marginal carbon during extreme cold events may spike for a few hours as more #2 jet peakers cover the shortfalls, but after June 1 2018 the demand response markets should be able to take care of most (or even all) of it.

    I'm not crazy about long haul transmission line solutions either, but it's probably the lesser of several possible evils. The New England Clean Power Link VT has already been approved by FERC and will probably get built, and would serve quite a bit of the marginal MA peak load. I'd expect the Maine Power Express under-sea connector from eastern MA to Maine to tap connectivity with New Brunswick will probably get built too, but I'm a bit more skeptical about the prospects for the Northern Pass, unless folks in New Hampshire support it more than the gas pipeline expansions. For environmentalists looking at the local ecosystem disruptions of each it's a bit like asking if you'd rather get stomped on the foot or kicked in the butt. From an atmospheric carbon point of view taking advantage of existing excess hydro power is clearly better than feeding the gas power plants (or building more gas powerplants when the fuel capacity increase makes existing nukes even less competitive relative to gas.)

    A near ideal solution to wintertime gas grid connection would be massive deployment of heat & power micro-cogenerators, but that would take a lot more time to install. Power output of cogeneration would rise & fall with heat load, and would apply a slice of the gas-grid-capacity pie currently reserved for space heating and put it onto the power grid at the appropriate peak times for heat pumps and resistance space heaters.

  34. ElectricHoosier | | #34

    Heat Pumps Keep Getting Better
    We've been successfully using a split ASHP since 2001 and I've noticed significant technological progress in the last 5-10 years in the USA/Canada. The HSPF example of 11.3 is OK, but class leading is now 13+. For reference, our '01 model is HSPF 8.5 in zone 4 and still going strong with a scroll compressor. The average gas furnace being used isn't 95% efficient either.

    Copeland has finally announced V2 of their variable speed scroll compressor at 25+ SEER, 13 HSPF, and variable capacity now down to 20% of nameplate. Hopefully they've also improved the cold weather noise issue and offer it in 1-2 ton models for tighter homes. This mainstream progress will drive significantly higher adoption of heat pumps in many areas, assuming they price it right. Mini-splits are great, but wider adoption will require better split units to fit the housing stock.

    Other recent improvements are CO2 HPs, reverse cycle chillers for homes, motel/hotel heat pump PTACs, and better HPWHs. Add to this Wi-Fi thermostats that tell you how much energy you use and differentiate between heat pump and backup electric resistance use. So now you have a vi$ual incentive to turn off the backup strips. It took me a long time before I realized how to go into hidden settings in our old Carrier unit and turn off the wasteful strips. There was no guidance from the HVAC techs when the HP was installed. To their unintentional credit, the 2.5 tons of capacity provides just enough heating capacity in S. IN with the heat pump alone for 99.9% of winter hours. Design temp. is 9F, but strips won't kick in until -3F, and then only for a short while. Home is circa 1982 bi-level and moderately tight.

    Lots of MISO territory wind energy is available on winter nights in the Mid-West to power heat pumps and displace old coal energy! It would be interesting to see actual marginal carbon production.

Log in or create an account to post a comment.



Recent Questions and Replies

  • |
  • |
  • |
  • |