Welcome back to our series of guides covering how to power your life with renewable electricity. If you’re new to this series, check out the introduction in Part 1.
Articles in the series:
- Part 1: LED lighting
- Part 2: Water heating
- Part 3: Transportation
- Part 4: Home heating and cooling
- Part 5: Solar panels
- Part 6: Cooking
- Part 7: Laundry
Part 4: Home Heating and Cooling
What is an air-source heat pump?
Just as a refrigerator can pull heat out of a freezer till it’s below 32°F, an air-source heat pump can pull heat out of outside air even when it’s below freezing and deposit that heat in your house.
But don’t house heat pumps stop working below 40°F outside temperature? That used to be true, but there are now “Cold Climate Air Source Heat Pumps” that work efficiently down to 5°F on average and some work as low as -30°F!
The picture to the right shows the heat pump in my office heated to 117.9°F when it was 35°F outside. That temperature comes purely from pumping heat out of 35°F air.
Even better, heat pumps can pump heat in either direction, so one system can act as both a heater and as an air conditioner. In fact, a traditional air conditioner is basically an air-source heat pump that is designed to only move heat from inside to outside.
Ground-source heat pump
Heat pumps can move heat between various places. Moving heat from deep underground where it stays a similar temperature all year is a lovely idea, but it costs a lot to set up. A GreenBuildingAdvisor article written in 2013 found that ground-source heat pumps might only make economic sense in the coldest climates, and only because of a 30% federal tax rebate, and only if you had experienced and affordable local installers available. Sadly, that 30% credit ended at the end of 2016 and I highly doubt it will be coming back. On the other hand, some states may still offer good credits.
Without a credit, I can’t see how ground source could make sense for many situations. Specialized air-source heat pumps now work down to -30°F, so there are no longer climates where ground source is required unless you live north of the arctic circle. As air source continues to dominate the market, it will continue to pull ahead of ground source in terms of price reductions, reliability improvements, and good installers.
Air-to-water heat pump
An air-to-water heat pump moves heat from outside air into water. According to GreenBuildingAdvisor, air-to-water heat pumps:
- Are more efficient than any other method for home heating, but also more expensive.
- Use heated water run through pipes to radiators throughout your house.
- Are most economical if installed during home construction when it’s cheaper to run water pipes through walls.
- Can double as your hot water heater as well as home heater.
- Dominate the market in areas where air conditioning isn’t necessary, such as Northern Europe.
- Can theoretically be used for home cooling, but water pipes need careful insulation to eliminate condensation that would cause mold growth. You would also need to mix the water with antifreeze which would make it unsuitable for heating hot water for your faucets.
Basically, if you’re building a new house in an area that doesn’t need air conditioning, you might look at this option. Otherwise, stick with air-to-air heat pumps. We covered various models of AWHPs for heating hot water in Part 3, and you could use the exact same units for home heating.
Air-to-air ductless mini-split heat pump
An air-to-air heat pump moves heat from air outside your home to air inside your home. Many Americans will immediately assume that heat is moved to a large central unit from which a series of air ducts blow the hot or cold air throughout the house. However, doing it that way is the least efficient way of moving that heat.
A “ductless mini-split heat pump” doesn’t blow air through ducts, it pumps hot and cold using refrigerant lines. Those lines are the same copper tubes that you see in air conditioners and refrigerators, just longer. While blowing hot and cold air through ducts can waste up to 30% of the hot or cold you were trying to deliver, insulated copper tubes waste only about 5% of the energy before delivering it to the exact room(s) you want conditioned.
That per-room delivery also saves energy. We keep just our bedroom warm at night, but the central gas heater previously leaked warm air all over the house at night even with just the bedroom vent open. It’s also very nice to automatically heat the bedroom just before we usually sleep rather than flipping vents open and closed all over the house every morning and every night, then waiting for the bedroom to heat up while we shiver.
Central air ducts are difficult to size correctly, especially in areas that require heating and cooling. Our air ducts were sized for the gas heater when the house was built, which meant the AC that was added later could not deliver enough air upstairs where we needed cooling. Even with all vents downstairs closed and a strong fan blowing air up the stairs, downstairs would end up freezing and upstairs might drop 1 degree every few hours at most. With half the vents closed, not enough air circulated through the system, so the AC refrigerant would occasionally freeze, requiring us to stop the system, wait awhile, and turn it on again. Ductless systems avoid all these issues.
“The United States is one of the few world regions where ductless systems are not the standard product type for residential cooling.”
In US areas where natural gas lines don’t run, such as cold New England, ductless systems are becoming popular for their efficiency. Out in sunny California, where mini-splits would save enormously vs ducted air conditioners, the first guy I talked to said he’d only installed one system and it was intended for heating. I guess people really get stuck in their ways.
What do mini-splits look like?
A mini-split “head” sits at the end of copper refrigerant lines and converts the heat or cold they contain into hot or cold air that blows into the room. Our largest “head” is in the living room in the left picture (it also heats the kitchen), while the smallest head is in the bedroom:
Yes, I know there are more dragons than heat pumps in those pictures.
The insulated refrigerant lines, electrical power, and indoor condensation drain lines all run through 3″ diameter holes in the wall out to these covered tracks outside the house, converging on one main condenser that’s similar to the outside unit of an air conditioner:
When the indoor heads run in cooling mode, water condenses on their cold interior parts. That water is collected in a drain line, and those drain lines run down below the condenser to drip away from the house.
When the indoor heads run in heating mode, water condenses on the cold outdoor unit. That water just drips onto the deck, so we plan to add a pan under it, though we’re just using a mulch bin for now.
In colder climates, it’s important to have the outdoor unit under an overhang and lifted above the ground to prevent snow piling up high enough to interfere with its operation. It will melt snow automatically, but it wastes energy to do that.
Mini-split heat pump price
Overall, you can expect to pay around $2000 per mini-split “head” for U.S. installations (that includes hardware and labor). I’ve heard it’s significantly cheaper in areas where mini-splits are more common. You don’t necessarily need one head per room, but conditioning only the rooms you occupy can save a lot of energy in homes that aren’t super insulated. Super insulated homes often use one head in a hallway and open and close doors to let the air flow into the rooms that need conditioning. Interestingly, our unheated downstairs has been up to 20°F colder than our heated upstairs, so air between floors doesn’t mix as much as you might expect even with an open stairway linking them.
Considering that heat pumps work to both heat and cool, their cost should generally be less than a new AC plus a new furnace. The problem with that bit of cost saving is that few people need to replace both AC and furnace at the same time. In new construction, I see little excuse not to go with mini-split.
I usually research the heck out of large purchases but I was distracted with our solar system installation so when our installer said Mitsubishi equipment was the only mini-split option available from his supplier, I just went with it. Mitsubishi systems do seem to get the most press, but other popular brands are LG, Fujitsu, and Daikin. Check out this comparison.
I would avoid buying any of the “bargain” mini-split systems, as you will probably pay more in the long term for repairs. Interestingly, Mitsubishi sells a lower-cost mini-split through Home Depot, but our installer said it had a low-quality look to it.
If you live in a region that stays below freezing for long periods, definitely choose one of the Cold Climate Air Source Heat Pump models. They’re more expensive but will pay for themselves in energy savings.
In Part 3, we talked about the efficiency of CO2-based heat pumps. Unfortunately, I could find no CO2 air-to-air heat pumps. In fact, it looks like CO2 units are all air-to-water or water-to-air. CO2 beats other refrigerants when pulling heat out of very low temperatures, but it’s 10–15% less efficient in air conditioning applications, which may explain the lack of air-to-air CO2 heat pumps.
Although R410A, a common mini-split refrigerant, is being phased out by Montreal Protocol rules, it won’t be completely banned till at least 2030. So far, companies seem more prone to replacing R410A with R32 instead of CO2, so I don’t hold a lot of hope for CO2 mini-splits to appear in the near future.
Mitsubishi currently offers R32 air conditioners in Japan, Australia, and the UK sometime in 2017. R32 causes less than a third as much global warming as R410A and is 6% more energy efficient. I can’t find anyone talking about installing Mitsubishi R32 mini-splits in America, but if Mitsubishi won’t sell one to your installer, Daikin, Panasonic, and Fujitsu are supposedly ahead in releasing them to various markets.
Our experience with Mitsubishi Mr. Slim
The Mitsubishi “Mr. Slim” system we installed is amazingly quiet. We have its main compressor (model MXZ-4C36NA) installed ~7 feet away from our bedroom wall but it’s hard to tell it’s on from in the bedroom. On the other hand, the vibration of the main compressor does transfer through the wall the compressor is mounted on into the room behind that wall, so keep that in mind. I can actually feel the vibration faintly when close to the wall, and I can hear it running, but not very loudly.
We have four indoor heads connected to the outdoor compressor. The indoor units are not as silent as ducted vents, but not loud enough that I notice them unless I’m listening. Although they look a little like hotel air conditioners, heat pump heads contain only a fan, not a compressor, so there’s no chance of them ever developing the kind of vibration you sometimes hear in hotel rooms. The most notable sound the head in my office makes is a crackle that almost sounds like rain hitting a window. This seems to only happen for about 20 seconds soon after it turns off at 11pm, and only when it’s freezing outside, so it has something to do with temperature expansion/contraction. If the softest noise bothers you at night, you can get a ducted head for the bedroom that will be a little less efficient but keep almost all sounds muted outside the bedroom.
Traditional HVAC equipment uses single speed or dual speed motors that put out just one or two levels of hot/cold. They must turn on and off frequently to keep a stable temperature in the room. Mr. Slim uses inductive motors to run at whatever speed is necessary to keep the room temperature stable, and that’s more energy efficient and less prone to creating cycling sounds that might disturb you.
My main complaint with Mr. Slim is the included remote controls are terrible. Stock remotes can’t even be programmed to turn each head on/off at a particular time each day. To get any decent features, they expect you to buy an “optional” thermostat for $260. I was able to get the same thermostat for $150–$199 on ebay but that still feels overpriced and it took many weeks to find four of them under $200.
On the plus side, all four thermostats I got are accurate (unlike many cheaper thermostats I’ve bought in the past) and they have the most sensible software I’ve seen on a thermostat. For example, rather than forcing you to set a particular temperature for all four time periods, you can disable up to 3 of the periods. You can also set a period to turn the system completely off. Tapping up or down arrows temporarily overrides the schedule and shows an unambiguous “Hold until: <time>” where <time> is the next regularly scheduled change. There’s also options to hold forever or set a time to turn off permanently if you plan to leave. I also bought a RedLINK Internet Gateway to control the whole system online in case we forget to set the temperature low before leaving on a trip.
Instead of buying the expensive thermostats, I tried a cheap Solrus SR507 remote that let me turn each head on/off at a particular time on weekdays and a different time on weekends. One problem is the head beeps when turned on/off via this remote and that can wake you up in a bedroom. A bigger problem is that without an external thermostat, the temperature sensor in the head is used, but that sensor is too close to the source of the hot/cold so you end up heating or cooling the room too far. We definitely felt that overshoot before we got the external thermostats. Using the head temperature can also make the outdoor compressor rev up and down too frequently as it chases the rapidly-changing head temperature and that reduces efficiency. It’s a major failing that Mitsubishi doesn’t include the external thermostat with each head or provide a remote with a temperature sensor in it like Fujitsu does.
Sizing a heat pump
- room sizes
- area of walls
- sun exposure of each wall
- size of windows and doors
- overhang of roof
- insulation ratings
- and more.
Although building codes generally require systems to be sized according to Manual J, GreenBuildingAdvisor says it is often difficult to get an HVAC company to perform a Manual J and also difficult to get them to give you the results. “As with many code provisions, the requirement for Manual J and Manual D calculations is widely ignored and rarely enforced.”
Because Manual J requires so many measurements, HVAC installers would rather use a simple rule of thumb like “you need 1 ton of cooling per 500 sqft”, but those estimates can be highly inaccurate. Energy-efficient homes built to higher insulation standards need drastically less heating/cooling than older homes, and Manual J accounts for that.
If you’ve got a home that’s average for your area and you find an installer you trust that seems to have done so many installs they “just know” what you need, that’s probably fine. If your home is unusual, you should probably insist on finding an installer that’s both willing to do a Manual J plus a Manual S to size the system. Beware: installers often want to go bigger than Manual S suggests “just in case” even though the calculations already include a “just in case” factor.
I love knowing how things work, so I found an online Manual J calculator and spent many hours measuring things to fill in every field it asked for. Even so, I had to put average values for things I couldn’t measure, such as construction tightness and duct insulation. Since we’re using a ductless system, I’m not even sure what to put for duct insulation, but I assumed “conditioned”, “R-8”, and “very tight” was best. Construction tightness controls how much air seeps in/out through the walls of your house. The only way to get a really accurate measurement is to blow air into the house through a fan sealed in an opening and measure how much the air pressure increases (lower increase means the house is leaking more air).
Doing Manual J correctly is complicated, time consuming, and even if you’re willing to take the time to learn how to do it, measuring tightness is something few of us have the equipment to do. I recommend hiring someone to do it right. Just make sure they actually do it right: they should be measuring every room, every window, every entry door, crawling in the basement and attic to look at insulation, and running that fan pressure test. It may cost a bit to do Manual J, but may save even more if you avoid installing an oversized system. HVAC installers will always go oversized if they use a rule of thumb estimate because profit margins are higher and they don’t have to worry about angry customers calling saying they’re uncomfortable due to an undersized system.
Manual S is much simpler than Manual J, but it’s a bit tricky with a heat pump. I recommend letting an experienced installer do your MS calc but if you want to double check their work, this article describes the process. Our Manual J calc said we needed 31.4kBTU of heat on the coldest anticipated day so our installer selected a 36kBTU heat pump. That would have been fine with a gas furnace because it produces 36kBTU regardless of outside temperature. On the other hand, our heat pump reduces to 68% of its optimal heating capacity when it’s 28°F outside. 0.68 * 36kBTU = 24.48kBTU, not nearly enough to keep up with our 31.4kBTU of heat required.
Note that 68% heating capacity at 28°F figure comes from a graph in the service manual for our heat pump which I managed to find online. All heat pumps reduce the amount of heat they can generate as outdoor temperatures fall, but each model does so at a different rate.
Luckily, Manual J assumes you’re heating the whole house to the “design indoor temperature” (usually 70°F), but we never heat the whole house. The coldest part of most days is at night, and we only heat our bedroom at night, so the heat pump has no problem keeping up. We only had a problem on two days in January that were dark, rainy, windy, and around 34°F outside. We let the upstairs cool overnight, and in the morning the heat pump was only raising the temperature by 1 to 2 degrees every hour. At 11am it was still 5 degrees below our set point in the living room so we set up some electric space heaters to bump it up quickly. Having to do that a couple days a year isn’t the end of the world, but it does waste power compared to sizing the heat pump correctly.
Smaller size systems also mean slower heating. Our 80kBTU gas furnace is so oversized for our house that we can heat our largest rooms from shivering to sweating in an hour or less. The heat pump heads, rated 9kBTU in small rooms and 18kBTU in large rooms, don’t have the capacity to raise temperatures that quickly. In our 352 square foot downstairs room that gets no sun, the 12kBTU heat pump head once took 6 hours to raise the room from 54°F to 65°F with an outside temperature of 38°F and falling. If we used the room daily, we could turn on the heat early and it would be fine. Unfortunately, it’s a room we use infrequently, but when we want it, we don’t want to wait hours for it to heat. Definitely talk to your installer if you have a room like that.
If you don’t mind spending money on a larger system to heat faster, a bit of oversize won’t usually hurt efficiency on a variable-speed heat pump like Mr Slim. Traditional furnaces only operate at one speed, so to hold a temperature it may cycle on/off frequently, which reduces efficiency. Since Mr. Slim can operate at a wide range of speeds, ours can maintain top efficiency operating from 1 to 3 ton capacity.
Mitsubishi heat pump thermostats actually learn how long it takes to raise the temperature in each room so if you set it to a temperature at a certain time, the thermostat will start raising temperature early to get it where you want it at the correct time. It doesn’t always make it on the coldest days, but it’s rarely been off enough to notice. Unfortunately, that behavior doesn’t work if you set the pump to turn off completely, say overnight. Instead, you have to set it to a low temperature overnight (50°F is the minimum).
At our house, the sun helps heat the top floor, so the heat pumps let us raise the temperature upstairs much faster than downstairs. The sun may be less of a heating benefit in colder climates where houses have better insulation or where the sun is out on fewer days, but it’s something to keep in mind. I don’t think Manual J assumes any solar heating because some days are cloudy and the system has to be able to keep up on dark days. The fact that warm air rises also helps heat the upper floor faster – it’s always warmer upstairs even with no sun and no heating. There are a lot of things you have to take into account, but a good HVAC installer should be able to offer advice on system size in each room.
Choosing an installer
Even if mini-splits are rare in your area, they use all the same parts as a traditional air conditioner so any AC tech should be able to repair a mini-split. The problem is that in areas where mini splits are rare, you will encounter a lot of installers that don’t know what they’re doing. I had one guy tell me heat pumps don’t work below 40°F (true years ago, false now) so he tried to discourage my installing one. Depending on your area, it may take some patience to find a qualified installer. I definitely recommend finding someone with experience in installing heat pumps with multiple heads.
Yelp.com has always been my best source of reviews for all sorts of things, including HVAC installers. Google Maps is also helpful and often links you to reviews from other sites. I was recently pointed to Buildzoom.com, which was created specifically to search for home-improvement workers and provide feedback. I haven’t tried using them yet but they look promising. If you can find reviews that mention a successful multi-head heat pump install, bump them to the top of your list!
Beware of recommendations from friends that are only as good as the research the friend put in to picking the installer. My neighbor picked his solar installer based solely on a recommendation from another friend, then recommended that same guy to me, but the guy knew less than me about every question I asked, charged 33% more than industry average, and took a lot longer on my neighbor’s project than the group I went with.
Two summers ago, when I tired of sweating during a long heat wave, I called up a well-reviewed HVAC company to repair our rarely-used air conditioner. They were so busy that they referred me to a semi-retired guy by the name of Jim. I got along well with Jim and he didn’t mind explaining things or even letting me work on a couple of the easy (but time-consuming) bits of AC repair to save a little money. When it came time to install the mini-split system, he was my first choice and he did a good job even though it was his first time installing such a large system. Unfortunately, his lack of experience with larger systems means we’re stuck with the undersizing problems I mentioned earlier, so he might not have been the best choice.
With natural gas prices at historic lows when I started my heat pump research, the payback period on our heat pump system when linked to free solar energy is a dismal 32 years. This article predicts natural gas prices to double and remain around their historical average, in which case the payback period is 16 years. In fact, natural gas futures are now showing prices have already returned to their historic average. Either way, with a 20 to 30 year PACE loan, you won’t be paying enormously more than you currently pay for gas heating, and you may pay less.
If you heat using fuel oil, propane, or even wood, the payback period should be much shorter.
To check for rebates in your area, click here, then use the Apply Filter button to narrow down results to your state.
Add a filter for Coverage Area > County or City or Zip Code (although the way they force you to choose a zip code from a list is really frustratingly slow in California due to the number of zip codes).
Next, add a filter for Technology > Energy Efficiency > HVAC > Heat Pumps.
In my area, I picked up a $300 tax rebate. Unfortunately, that particular rebate is not available for systems installed after 2016.
Estimating energy use
Estimating how much energy a heat pump will use is tricky because its efficiency changes based on the temperature difference between inside and outside, and each brand/model has a different efficiency at each temperature. On the other hand, installing a heat pump and waiting until you’ve gone through a summer and/or winter to see how much energy it actually uses can greatly delay your solar installation and use grid electricity that will emit more CO2 (and cost more) than running a fossil heater. So, it makes sense to get a good energy estimate as described in these indented sections:
Find the energy you currently use to heat your home
If you use a natural gas heater, look online and it should tell you how many therms you used each month for the last year. If there is no monthly history link, you may need to download a copy of your monthly bill for each of the last 12 months.
Find your average monthly use during summer months when you aren’t using gas heating. This should roughly equal the gas you use for everything but home heating each month.
Multiply your monthly summer average times 12 and subtract that from the total number of therms you used in the year. What’s left should be therms used for heating.
I repeated the above calculation for 3 years to get a better average, though that’s mostly because our gas company regularly neglects to read our meter so the usage can bunch up in the wrong months when they finally read it. We’ve been forced to pay too much when they do that because they throw most of the gas we used for a few months into one month, then charge a penalty for using too much gas that month. This scam has been reported by many people, including my father, but Southern California Gas (part of Sempra Energy) somehow keeps getting away with these overcharges and made no corrections when either of us complained. This is the same company that set up a fake gas supply shortage to pressure California regulators to re-open Aliso Canyon after the most environmentally damaging gas leak in American history was caused by failing to replace a safety valve removed in 1979.
I’ll be very glad when I ditch Sempra Energy.
Our average winter gas heating use is 285 therms for the year. Later on, we’re going to need to convert that to British Thermal Units (BTU), so let’s do that now:
285 therms * 99976.1 BTU per therm = 28,493,188.5 BTU.
Here are some conversions for other fuel sources:
1 gallon of fuel oil = 43.9kWh electricity = 149,793 BTU
1 gallon of propane = 27 kWh electricity = 92,127.8 BTU
1 kWh of electric resistance heating = 3412.14 BTU
But wait, gas appliances aren’t 100% efficient. If you have a modern furnace, you can probably look up its efficiency by model number. On an old furnace, you’ll need to turn it off (unplug it, or turn off the circuit breaker that powers it), then open up some access panels and look inside. You should find a plaque that shows BTU input and maybe BTU output:
On our 1980 furnace, the plaque shows input is 95,000 BTU/hr but it doesn’t show output. I searched Google using the model number for “397HAW048080 btu output” but got nothing. Searching for “397HAW048080 manual” eventually got me the manual. This type of furnace is made by Carrier (now merged with Bryant) and usually referred to as “397H series” or “397HAW”. The 080 at the end of the model number means its BTU output is 80 mega BTU, aka 80,000 BTU/hr. All this means it burns 95,000 BTU of gas but only generates 80,000 BTU of heat. 80,000 / 95,000 = 0.8421, so the furnace is 84.21% efficient.
28,493,188.5 BTU * 0.8421 efficiency = 23,994,114 BTU per year in energy needed to heat our house with 100% efficiency.
If you’re using fuel oil, propane, or some other fossil fuel to heat, you will need to research the efficiency of your heater. Most heaters seem to hit around 85% efficiency.
Use historic BTU usage to estimate heat pump energy
Heat pumps are rated using Heating Seasonal Performance Factor (HSPF) to estimate their electricity use. Higher HSPF values are more efficient. HSPF even compensates for temperature changes throughout winter to give an average energy use over the heating season, but it’s based on a particular climate (zone IV) where not all of us live. A joint study by Florida Solar Energy Center and Berkley Solar Group goes into great detail about how HSPF values translate into actual energy use in the 15 different climate zones shown in the graph. Skipping past all the details, this equation is all we need:
0.1041 – 0.008862 * T – 0.0001153 * T ^ 2 + 0.02817 * HSPF = % decrease
The above formula will work with an HSPF 8.5 system or above. HSPF above 10 is uncommon, though Mitsubishi makes a “Hyper-Heating” line of heat pumps that go up to HSPF 13.5 and work down to -13°F. You shouldn’t need to go below HSFP 8.5 unless you’re planning to buy a cheap and inefficient system that may have maintenance problems, but if you need to go below 8.5, use this formula instead: 0.1392 – 0.008460 * T – 0.0001074 * T ^ 2 + 0.02280 * HSPF
T in the formula is the “99% Design Temperature” which is simply the temperature, in Fahrenheit, that the area around your home has stayed above 99% of the time over the last 30 years. To find T:
Check this table to find your nearest city.
The value in the Heating DB > 99% column will be your T value, but it may need adjusting.
Compare the Elev column to your home’s elevation. If you don’t know your elevation, look it up here. You can also find elevation using Google Maps by getting directions between two points that are right next to each other (ie your house and your neighbor), then click the bike rider icon and a little elevation map will appear.
If your home is higher than the Elev in the chart, subtract 1°F from your T value for every 250 feet of difference (i.e. subtract 4°F for a 1000 foot difference). If your home is lower than the Elev in the chart, add 1°F for every 250 feet of difference. Temperature changes by elevation vary depending on the humidity in the air, so 4°F per 1000 feet is an approximation. With high humidity, temperature may change as little as 3.3°F, but with no humidity it can change as much as 5.4°F. Use 5.4°F if you live in a wet area, or 3.3°F in a dry area.
In our case, the nearest station has a 99% design T of 37.2°F and is located 3862 feet below us, so I adjust my T value by subtracting 4°F * 3.862 = 15.5°F. My final T value is 37.2 – 15.5 = 21.7°F.
Plug that T value into the formula above and I get 0.1041 – 0.008862 * 21.7 – 0.0001153 * 21.7 ^ 2 + 0.02817 * 10 = 0.139 = % decrease.
This result tells us a heat pump will work 13.9% less efficiently than its HSPF rating in our climate. The study says % decrease should be used to adjust the HSPF rating using the formula:
HSPF * (1 – % decrease)
Plugging our HSPF 10 heat pump into the formula, we get: 10 HSPF * (1 – 0.139) = 8.61 HSPF. In other words, our heat pump behaves like an HSPF 8.61 system in our climate.
An HSPF 8.61 system moves 8.61 BTUs of heat into the home for every watt-hour of electricity used. We need 23,994,114 BTU per year of heat, and an HSPF 8.61 heat pump can give that to us for 23,994,114 / 8.61 = 2,786,772Wh / 1000 = 2787kWh per year.
Add 26% if your home is not super insulated
By super insulated, I mean a home built to ENERGY STAR or “net zero” standards. I’ll discuss in detail why I think adding 26% is necessary later on, but basically we seem to be using about 26% more energy than the above estimate and I think it’s mainly because our insulation is average for 1985. Even if your home is above average in insulation (but not super insulated) I still recommend adding 10–20% to make sure you don’t underestimate your heating energy requirements.
2787kWh/yr * 1.26 = 3512kWh/yr
Add 6.5% to heating estimate to account for grid losses
Home heating accounts for the largest use of electricity that can’t be timed to occur when solar is producing power. The greatest need for heating is after the sun goes down. So, most of our heating electricity will come from the grid, and some of that power will be lost during transmission.
From here: Average transmission/distribution losses for the USA in 2007 were 6.5% (and 6.6% in 1997). That means electricity used to heat at night will take around 6.5% extra energy to reach you, thus you need to supply 6.5% more power during the day to offset that. On the other hand, the excess solar power you supply to the grid during the day should be completely used by houses around you and thus should have negligible transmission losses.
3512kWh/yr * 1.065 = 3740kWh/yr.
Find the energy you currently use to cool your home
Assuming you use significant electricity for summer air conditioning, estimate that electricity as follows:
Log on to your electric company’s web site and download monthly electric history for the last year.
Find the average amount of electricity you use during a few months where you use no air conditioning or electric heating. Call this average A. Note that even gas heaters use a fair amount of electricity to run fans and burner starters.
Add together the total electricity used in all months where you use air conditioning and call it B.
If you use extra electricity for anything during warm months that you don’t use during cold months, you need to find a way to subtract that extra electricity from B. (i.e., you might run a pool filter only in the summer and you need to subtract that from B.)
Multiply A by the number of months you use air conditioning, then subtract that value from B to get the total kWh of electricity you use for air conditioning each year. Call this total C.
Estimate heat pump energy needed to cool your home
Both heat pump and air conditioner efficiency are measured using a Seasonal Energy Efficiency Ratio (SEER) rating. A system with a SEER rating of X moves X BTU of heat out of your home per watt-hour of electricity used. Like HSPF, SEER is averaged over the cooling season for a particular climate.
You can use the SEER rating of an existing air conditioner and the SEER rating of a more efficient heat pump to adjust the C value calculated in the previous section as follows:
C * (AC SEER / Heat Pump SEER) = Heat pump kWh/mo
The study I linked to earlier also gives us a formula to adjust the SEER rating to be more accurate for your particular climate. The only reason you might need to use the formula is if you don’t have an AC but you do have some idea of your cooling load from doing a Manual J calculation discussed earlier. I’m not going to go into detail but the math will be very similar to the heating calculations I described earlier. Use this formula for SEER values of 13.5 or greater: -0.5864 + 0.005668 * T + 0.01029 * SEER = % decrease. For SEER values under 13.5, use this formula: -0.5655 + 0.005414 * T + 0.01039 * SEER. Your T value will be found on the left side of the Cooling DB/MCWB > 1% > DB column in this table.
Since we rarely used our air conditioner, I didn’t bother trying to estimate how much energy the heat pump might use for cooling. I figure if the solar panels are producing a lot more power than we need I’ll use some of that excess for cooling. Otherwise, opening windows and doors plus using a fan works for all but a few unusually hot days here.
If you live in an area where air conditioning is the primary use of the heat pump and your home is not super insulated, you should probably add 26% to your estimate.
Does this estimation fit reality?
I bought The Energy Detective (TED) and delayed this article so I could get some actual data on our heat pump’s energy use. I originally tried to estimate heat pump energy using total power use reported by SCE plus solar generation minus an average day’s electricity use before the heat pump. This showed the heat pump using significantly more power than expected. I installed TED on Dec 8th and soon learned the biggest problem was the SolarEdge monitor reporting about 6% more power generated than was actually generated. To double check, I used a Uni-T UT210E clamp amp meter to verify TED was accurate. It’s a cheap meter, but I verified it matches the output of an expensive Fluke meter my heat pump installer was using.
After adjusting for the incorrect solar reporting, I painstakingly went through day by day and hour by hour subtracting anything that appeared to be car charging or other above-average power use to estimate heat pump use in Oct and Nov. After Dec 8, I had actual data from TED. Here are the results so far:
The rightmost Heat pump energy use column is based on TED data from Dec 8th through Mar 31st, and it comes fairly close to matching the Estimated heat pump based on gas use column. That estimated column uses the expected average yearly efficiency of the heat pump, but efficiency gets worse as outdoor temperatures drop. Thus, it isn’t surprising that in the coldest months of Dec and Jan, we use more energy than the estimate, while in warmer Oct–Nov, and Feb–Mar we use less energy than the estimate.
The grey values in Apr–May are based on the actual Mar energy use weighted by historic gas use in each month. I don’t expect to use heat pump heating in June-Sept, so assuming I come close to those grey values, total Heat pump energy use will be 2285kWh/yr, well under our original estimate of 2784kWh/yr (before I added 26% and another 6.5% to cover transmission losses).
So it seems like we’ve overestimated heating energy, but there’s one big problem:
We have a downstairs room that got heated a lot more in the last few years than it did this year. My best estimate of how much power we’ve saved by not heating that room is 1224kWh.
2285kWh main rooms heating + 1224kWh downstairs room heating = 3509kWh/yr, and that’s 26% higher than the original 2784kWh/yr estimate.
How sure am I of the 1224kWh estimate?
Since Dec 14th, I’ve logged the outside and room temperature every 15 minutes using a Raspberry Pi with a temperature sensor. Unfortunately we only heated the room two days in December and two in March, so that didn’t give me enough data to fit those outside and room temperatures to an energy estimate with high confidence. I did tests on 3 more days in early April overnight when temperatures were 55 to 67 to flesh out the graphs on the right.
The lines fitted to the points changed significantly as I added points, but the overall energy estimate only increased 4.25%, so I don’t think adding more points will greatly alter the estimate. Even if the estimate were somehow 25% off, that only shifts the error in the total energy estimate from 26% to 37%. So if you’re really paranoid about installing enough solar to cover your heating, add 37% instead of 26% to your heating estimate and you should definitely be covered.
Having to add 26% to the energy the manufacturer claims the system will use is a fairly large discrepancy, so I investigated whether or not my experience is unusual.
Studies of heat pump efficiency are usually measured using Coefficient of Performance (COP) instead of HSPF. A COP of 2.0 means the heat pump is providing twice as much heat as the electricity it’s using. Electric resistance heaters always have a COP of 1.0 by directly converting their electricity into heat. We can find the COP of our system by dividing gas energy use by estimated heat pump energy use:
- We used around 285 gas therms per year for heating.
- Convert 285 therms to electricity equivalent: 285 therms * 29.3kWh/therm * 0.8421 furnace efficiency = 7032kWh/yr.
- 7020kWh/yr / 3509kWh/yr = 2.0 COP.
2.0 COP is much worse than even the worst COP of 2.65 listed in the heat pump manual at 17°F outdoor temperature (our worst days were 22°F, and only at night). At 47°F outdoors, the manual says we should get 3.5 COP. Unfortunately, these numbers are all under ideal conditions such as having only one heat-pump head connected. Many things can reduce that efficiency, like adding more heads, more copper refrigerant tube, more bends in the tube, and running at max speed to recover temperatures each day (all of which we do).
The US Department of Energy funded a careful study conducted by James Williamson and Robb Aldrich from the Consortium of Advanced Residential Buildings (CARB) in Norwalk, Connecticut. They monitored heat pumps operating in 7 real homes and found real COP values of 1.1 to 2.3. That sounds like our experience. The homes were in areas colder than ours with average temperatures below 0°F where you can expect lower COPs, but they were also using some of the most efficient heat pumps for those low temperatures (ie Mr. Slim Hyperheat) which helps raise COP. They were surprised to find lower COP values than previous studies. Mr. Aldrich offered this possible explanation:
“Some studies use air flow rates published in the manufacturers’ literature plus the delta-T to calculate COP. I have big concerns about that method. The best method may be side-by-side co-heating tests in which the minisplit is compared to an electric-resistance heater. I tried to get funding from NYSERDA for that type of study, but so far we’ve been unable to get anyone interested.”
The study also found that small changes like setting fan speed to Low (for quietness?) instead of Auto could kill COP and that was likely responsible for their worst performers. They made a number of suggestions to keep COP higher:
- Choose a unit that isn’t oversized since minisplits that cycle frequently have a lower COP (we’re good here).
- Make sure that water, snow, and ice aren’t dripping on the outdoor unit (we’re good).
- Adjust the airflow louvers on the indoor unit so that they are slightly below the horizontal; this adjustment will maximize airflow (our fins are set to Auto which should be okay).
- Set the fan speed to high or auto, not low (we’re good).
- Don’t attempt thermostat setbacks (this may be our problem).
- Homeowners in cold climates should consider mounting the indoor unit near the floor to lower the return air temperature. While it’s possible to do this with any indoor unit, some manufacturers sell indoor units that are specifically designed for floor mounting. Another way to lower the return air temperature is to install the indoor unit in your basement.
Mounting near the ceiling is what you always see in literature. It’s probably more aesthetically pleasing in most areas of the home, and it leaves more space for furniture. With large air intakes on top and blowers out the bottom on most units, they are obviously intended for installs above eye level. Mitsubishi units even have an electronic eye that’s supposed to look down to see the room and direct heat towards people (seems like a gimmick to me). However, it’s also logical that if you primarily use the heat pumps for heating, floor mounting saves energy. I wish we’d known that but ours are all near the ceiling.
Should thermostat setbacks really be avoided?
“Don’t attempt thermostat setbacks” means don’t lower the temperature of rooms when you’re not in them. I’ve read various people online claim they use less power by leaving heat pumps on constantly.
The 4th comment at the bottom of this blog explains that if you set a room to get cold overnight and heat only in the morning, you are trying to draw heat from outside when it’s coldest outside, and you’re trying to heat at the maximum speed which is the least efficient mode of operation. If you kept that room warm all night, you would be pulling heat in when it was warmer (on average) outside and you’d be running at lower, more efficient compressor and fan speeds.
That explanation sounds logical, but my measurements of actual heat pump energy use in our home show it takes 2.19 times more power to keep our upstairs heated overnight vs turning the system off and warming upstairs just after dawn.
Why do my test results differ from what experts say? I suspect it comes down to insulation. In houses that are well insulated for long, cold winters, I assume leaving a heat pump on all the time would save energy.
Our house was built in 1985 and uses fiberglass R13 in walls and floors, R38 in ceiling, and double-pane windows (air filled). Caulking is poor to nonexistent on the corners of outside walls so we probably have a lot of air leakage in the walls and the rubber on our sliding glass doors is not keeping air from seeping through.
So, we could get a better COP by running our heat pumps all the time, but the total power used by doing that would be at least twice as high. If we could dramatically improve insulation we could probably keep things always heated and achieve high COP and low power use like this person. We’re going to start working towards that because adding ~1266kWh per year to heat an irregularly-used room will almost certainly push us beyond the amount of solar energy we generate per year.
Prevent pipe freezing
Our central gas furnace is installed in the basement and constantly emits some heat down there when it’s running. That wasted heat kept the basement well above freezing in winter, but without it, temperature monitors showed the basement coming dangerously close to freezing on a few days. I’m probably worrying too much because various sites like this say burst pipes rarely happen till under 20°F outside, which we never reach. Even if an area inside the house hits under 32°F it has to stay that low for awhile for pipes to freeze solid and create pressure buildup that can burst pipes. This is especially true if pipes have insulation around them, which ours do.
Nevertheless, I installed a thermostat-controlled outlet in the basement set to turn on a space heater at 32°F. If you live in an area that commonly gets colder than 20°F, you probably already have pipe insulation or even pipe heater cable wrapped around pipes. Even if you have such safeguards in place, if you plan to turn down the thermostat while you’re away, remember to turn off the water main and open all faucets to prevent any possibility of frozen pipe damage.
Next time — Solar!
Stay tuned for Part 5 where we’ll talk about installing the solar panels meant to provide power for all this stuff we’ve covered in Part 1 through 4. It will be an epic adventure!