First of all, thank you very much to all of you who contributed to this thread. I agree with all of you completely.☺ Remember, the reason Internet discussions are so acrimonious is because the stakes are so low…
Last week, I published an article titled “Minisplit Heat Pumps and Zero-Net-Energy Homes” on GBA. At the end of the article I asked readers to submit questions on topics that they’re looking to learn more about so I could provide a “mini-consultation” and answer their question while hopefully helping others with the same questions. Here are answers to the five questions that I picked.
Q. How do you decide when ducted minisplits (rather than ductless minisplits) are worth the time and expense to install them?
Answer 1 — I’d like to combine the questions included in Comments #1, #5, and #13. The essence of the question is: When do we have to transition from a point-source heating strategy to a strategy that provides each room with its own source of space conditioning? (I say space conditioning because we may ask this question just as well about providing cooling.)
The first question we should ask is, are doors open or doors closed? And a subset of “doors open” might be a small transfer fan moving air from a warmer space where a heater is located to a room without a heater. (Robb Aldrich at Steven Winter Associates has done great work on this strategy.) Let’s look at “doors closed” first.
Say a 120-square-foot room has contact area through the floor with a heated room. (Let’s assume that the room under consideration is above the heated room.) Let’s assume that the room has 150 square feet of wall area that faces the upstairs hallway, which we’ll assume is open enough to the first-level heated room that it is at the same temperature as that room.
Each of those areas is has a thermal resistance of about R-3. So there is a UA of (120+150)/3 = 90. This is a room with a pretty high area of contact with the heated room. For every °F difference across these assemblies, there is 90 BTU/hr of heat transferred. Say you want that room to be 68°F when the heated room is 71°F. The available heat transfer is 270 BTU/hr.
If the room to be heated has a 15-square-foot window that has a U-factor of 0.2 (a good triple-glazed window) and a wall area to outdoors of 150 square feet with a U-factor of 0.03, a ceiling of 120 square feet with a U of 0.02, and an air leakage rate of 4 cfm, then the room UA to outdoors is (15×0.2 + 150×0.03 + 120×0.02 + 4×1.08) = 14.2 BTU/hr.
If we divide 270 BTU/hr by 14.2 BTU/hr, we find that (when the room is unoccupied) there is an energy balance when the indoor/outdoor temperature difference is 19 F°. With a desired indoor temperature of 68°F, that’s 49°F outdoors. Not impressive — and this is a superinsulated house.
Now, add some internal gains and things change. Put a person in there with an iPad and maybe an LED reading light and you more than double the available heat, and now you have a balance at 45°F to 50°F outdoors. But if you’ve been gone all day, and it’s been 30°F outside, you’re not coming into a warm room if you left the door closed.
What about if it’s open? That helps a lot. How many cfm move through an open door? I think it’s different depending on where the door is. If it is up a floor, it has more air flow through it than if it is on the same level — the difference in height between the heated room and the door opening helps that movement of air and energy. I noticed this recently in a house with a minisplit in the main living space, and an open door to an adjacent bedroom, and open doors on the second floor to two other bedrooms. The first-floor bedroom was noticeably colder. It had less conductive contact area too, and more exterior area, so all things weren’t equal; but it was striking how much cooler it was. (The main level was 72°F, the upstairs rooms were 68°F, and the first floor room was 63-64°F.)
I tend to think of an open door as being equal to a decent bath fan — it’s worth 50-100 cfm. So at the low end, 3°F temperature difference buys you 150 BTU/hr, and at the upper end it’s double that.
So Aaron’s 3,000 BTU/hr bedrooms aren’t going to be heated by a point-source heater, at least to satisfy most people’s comfort criteria. My guideline is that if people will tolerate 4°F lower than the heated space (which in my mind means 72°F heated space, 68°F bedroom) and they leave the doors mostly open, then a point-source heater is viable when the heat loss is 1,000 BTU/hour and the room is occupied; 1,500 BTU/hr is kind of my soft cut-off for considering it. Beyond that, I’ll provide some electric resistance backup in those rooms.
A couple of other thoughts to confuse the issue:
- How hot is the heater? The plume of very hot air off a wood stove creates a significant layer of very warm air at the ceiling level that drives more conduction and more convection. A wall mounted minisplit is the opposite extreme (well, maybe a radiant floor is) because it mixes the air in the room and there is little stratification.
- Who is going to live there? In speculative development with multiple units, there will be someone who hates you because you didn’t put heat in their room. I speak from experience.
- Cooling is harder. A cooling source on the main floor won’t cool the upstairs.
Q. The thermostat for our Mitsubishi ducted system only goes down to 63°F. Do you know of any way to reset the thermostat to have a lower minimum temperature?
Answer 2 — This question is an easier one. In Comment #12, Eric asked about his four-year-old Mitsubishi multi-zone system and how to set the temperature lower than the 63°F limit. Today we’d use the controller Honeywell provides to Mitsubishi, the MHK, which allows communication over the Web, and much wider setpoints. I re-confirmed today with Mitsubishi engineers that this control is unfortunately not retrofittable to the older system.
So my suggestion is that you try just shutting the second floor ducted system down at night completely, and let the main floor wall cassette do the work overnight. This would work well when you’re away for a longer period of time.
Another way, perhaps a bit cumbersome, is to figure out a way to put a tiny bit of heat right at the thermostat and fool the sensor into thinking it’s warmer than it really is. We’re talking about two bulbs worth of Christmas lights… or mount your iPhone charger right below the thermostat overnight.
As to your question on the flexible duct work — flex duct has a bad rap because it’s easy to install so badly. In your case, where the runs are short, it virtually assures you that there is very little duct leakage, and it’s not long enough to cut down on the air flow unless it’s been installed such that it is crushed or otherwise improperly installed. I’d leave it, as long as the rooms being served are being adequately heated and cooled. (Disclaimer: I just put flex duct in my own system — using a 5 or 10 foot flex duct runout to the register boot makes the system quiet — just stretch it out and don’t crush it.)
Q. When your heating and cooling loads in a net-zero home are already smaller than the capacity of anything but a single wall-mount ductless minisplit, how do you balance correct sizing with adequate distribution?
Answer 3 — Leigha’s question in Comment #17 is a variation of the first question in a climate that will need cooling as well as heating. The house sounds like a single-story. With the bedrooms to the north in this very well insulated passive solar house, I’m assuming that they don’t have a ton of glazing, which in this case drives heat gains from the sun, as well as heat loss. So I might be comfortable using just a single unit in the main space if the occupants understood that the bedrooms will be cooler than the main space in winter and warmer in summer, and if the rooms met my 1,000 BTU/hr criterion mentioned above.
Having said that, the system you mentioned that incorporates a wall cassette for the main space, and a ducted unit for the bedrooms would be a step up in occupant satisfaction for many people, because the bedrooms are now able to have their own setpoint. A ducted unit with the temperature sensor built into the air handler (the stock setup) averages the return temperature from the bedrooms to decide how much heating or cooling to provide. You also have the opportunity to incorporate some good filtration in that system.
MARC’S ONLINE COURSE
Interested in learning more? Since 2012, I’ve been working with NESEA and HeatSpring to teach as part of with other experts from the NESEA community. Over 150 professionals have taken my Zero Net Energy Homes design course, and the next course starts on February 3rd. This course is an opportunity to study with me: to ask me questions for a full ten-week semester. You will walk away with a comprehensive understanding of all of the key components of a zero net energy home — envelope, systems, and renewables — and how they fit together, with key pitfalls to avoid, and numerical calculators for sizing peak heat loss, glazing amounts, annual energy use, and solar electric systems that will empower you to confidently design a zero-net-energy home. Successful students will actually do a full design of a zero-net-energy home, and earn NESEA’s Zero Net Energy Homes Professional Certificate. The course is approved for 12 AIA CEUs + 6 MA CSL credits (1 hour for Code, 1 hour for Workplace Safety, 1 hour for Business Practices, 3 hours for Energy).If you’d like to see some free content from the course, you can of my course here, or .
Here’s another approach: do the whole house as a single-zone ducted system. With a strongly passive solar design, the whole-house ducted system can serve to redistribute air from the warmer side of the house to the bedrooms.
As to limits to the length of ducts: these units don’t have the static pressure capability of traditional American central forced-air systems, so looking carefully at the unit specification is important, as well as good duct design. For example, Mitsubishi’s SEZ air handler (used on the Mr. Slim models) has a top setting of 0.20 inches of external static pressure — that’s low! The Fujitsu RLFC series has a capability of 0.36 inches on the smaller air handlers, and lower on the largest one in that series. So you have to check.
Q. Our current energy load calculations show around 50,000 BTU/hr for heating and cooling. We were considering forgoing cooling, but with the Daikin Altherma system we could possibly have it as an added bonus to domestic hot water and radiant floors. Unfortunately, the initial cost has us leaning towards a condensing gas water heater.
Answer 4 – Eric’s question in Comment #4 is about the economics of comparing a gas condensing water heater with a Daikin Altherma, which can also drive a radiant floor and make domestic hot water, and can provide chilled water for cooling in the summer. Before I address this, may I comment that a design load of 50,000 BTU/hour for a house measuring 2,100 square feet in Zone 5 seems way high — at 24 BTU/hr/sf, maybe even higher than a code-minimum home today. My quicky guideline that I tell people is to set a target of 10 BTU/hr/sf in Zone 5. So make sure your load has been properly calculated; and if it has, first spend your money on reducing that load.
It sounds as though the base solution of the gas water heater has no cooling, but even though the Altherma offers it, you don’t need to implement it, especially since it means a second distribution system — added to the quite costly concrete/steel floor hybrid, you’ll need a ducted system for the cooling.
To begin looking at costs and benefits, you need to compare fuel costs. If the gas available is natural gas at current fuel prices (let’s say $1.25/therm) and the heater operates at 90% efficiency, then the heat costs about $14/million BTU. Say your electricity costs are $0.15/kWh. To get the same cost per BTU, the Altherma needs to operate seasonally at a Coefficient of Performance of about 3 — three units of heating for every one unit of electricity used. I don’t have data on Altherma units other than a couple of retrofits with very extenuating circumstances, but from the published engineering data I think an annual COP of 3 in Zone 5 in a low temperature hydronic radiant system is plausible.
So at that point you’ve spent more in capital cost than the gas system for similar annual cost.
What the heat pump offers you is the possibility on making your own energy with onsite renewables, which you can’t get with the gas. Buying the renewables is a separate economic decision: What does electricity cost where you live, what’s the history of price inflation, what are incentives beyond the Federal 30% tax credit, etc.? The prices on solar electricity continue to drop, the systems are long-lived and reliable, and you’re investing in your own energy source. And even if you choose the gas heating system, you’ll still be using electricity, so that choice doesn’t preclude using onsite renewables; it just makes it hard to get all the way to net zero.
But please first work on getting that 50,000 BTU/hr down…
If it’s OK, I’m going to pull Peter’s question in Comment #9 into this, because he asks to compare minisplits with ground-source heat pumps (GSHPs). Peter, I approach this question in two different ways. The first is to recount some analysis done by the team who designed the Putney School’s 16,000-square-foot Net-Zero Field House. Being in southern Vermont, they felt that air-source heat pumps were perhaps on the edge of their applicability.
Fortunately, the HVAC engineers Kohler & Lewis from Keene, New Hampshire, had switched their office building to minisplits a couple of years earlier and had seen their system operate down to -15°F even though there was no published data at those temperatures. And then Andy Shapiro led the team through an interesting analysis. They asked, if the GSHP system could truly achieve a full point of COP higher than the minisplits (I can’t recall what values they picked for the two COPs, just that they were separated by 1!), what would the cost be of each heat pump system plus the cost of the solar electric system necessary to provide the energy consumed by the heat pump system? (In essence, the cost of the heating system cost plus its associated fuel cost over the years.)
If I recall correctly, the minisplits, even after conceding a higher COP to the GSHPs, came in something like $4/square foot less than the GSHP system and the smaller associated solar electric system. And as Dana said in his post, we now have some really excellent hard field data about the COP of minisplits in cold climates that is really encouraging.
The second way I approach your question is to say that one thing I really like about the minisplits is how they are packaged systems from a single supplier, and are highly engineered as a system and therefore very reliable. GSHP systems are, at least where I have practiced, essentially custom engineered and installed, usually by several entities who have a shared responsibility to make sure the systems perform. Knowing that this will cause howls of au contraire to arise, I will say that GSHPs have been the most problematic HVAC technology I’ve worked with, and so I choose the Japanese air-source equipment without a second thought at this point (and I have worked with GSHPs on projects ranging from zero-net-energy homes to buildings up to 70,000 square feet).
Q. How exactly are these companies determining the capacities of these units? Is there a way to actually calculate, or even estimate, how much heat these units will actually output at -13°F?
Answer 5 – On ratings and Rheannon’s question in Comment #14: the only way through this is to get published engineering heating capacity from the manufacturer and then vet it through their representatives. If there are contradictions, pursue them until you are satisfied or until you reach a point where your confidence is so shaky you abandon consideration of that product.
For example, Dana mentions the Daikin Quaternity, and yet I can’t seem to ever find a capacity table below 14°F. That’s not low enough to make me comfortable in specifying it. I know some Daikin VRF machines have been observed operating at -15°F, and I’ve seen the Mitsubishi Hyperheats operating at -22°F at a school in west central New Hampshire. In Zones 6 there are increasing numbers of these heat pumps being installed. (We did some Fujitsus north of the White Mountains this year.) In Zone 7, I’d be looking at some backup heat. It all depends on the client’s appetite for risk.
Tidbits on some of the rest of the posted questions
Machines that offer separate sensible cooling and humidity setpoints almost always will do that at a cost in efficiency, because to maintain a relative humidity setpoint without over-cooling a space they will use some form of reheating the air. This means they cool the air down to wring the moisture out, then heat it back up to avoid over-cooling.
They may do this cleverly — like using the rejected heat of the compressor for reheating — but this is not an efficient operating mode. It may be more efficient than operating a separate dehumidifier, but the two options would need to be carefully compared to make that determination. I know from experience that using the Dry mode in the Mitsubishi VRF equipment uses more energy than running that equipment in the Cooling mode.
I would tend to agree with Dana that most buildings with minisplits won’t need additional dehumidification, but there are always conditions where there is no cooling load but there is a moisture load, and if you can’t tolerate occasional excursions out of the ASHRAE comfort envelope (60% RH limit, if I recall correctly), then you need some independently settable humidity level. Scott, the best way to assess whether the minisplits work well in Dallas is to ask people who have been using them in similar applications to yours.
On point-of-use demand electric water heaters: Say you start with 45°F water and want a 1.5 gpm shower at 110ËšF — that load is just a smidgen under 15 kW, or 60 amps. So you need to consider electrical panel capacity first.
By comparison, a typical electric tank type water heater is 1/3 of that. You can buy the demand heater for about half the cost of a really well insulated electric tank type, and installation is probably a bit lower in cost. If you need two demand heaters, probably there aren’t savings over the tank type and no worries about trying to take a shower when someone starts the dishwasher or clothes washer or just turns the kitchen sink tap wide open.
When might a point-of-use electric be a reasonable choice? Small urban single-occupancy units where the laundry is in the basement. Basically, when only one person needs to decide which uses to satisfy with the single unit.
On EcoCute machines (Japanese air-to-water heat pumps): Yes, it would be great to get them into the U.S. I wonder if the barrier is a refrigerant that operates at 1,500 to 2,000 psi — What will UL say about that?
Marc Rosenbaum is the director of engineering at South Mountain Company on the island of Martha’s Vineyard in Massachusetts. He writes a blog called and teaches as part of the NESEA Building Energy Master Series, you can .