Why isn’t there a 3k BTUH cooling ductless minisplit?
rhl_
| Posted in General Questions on
Does any of the GBA pros have contacts at Mitsubishi/Trane to get some clarity on this?
It seems that it would be an easy way to make way more energy efficient installs for multisplit systems. There is obviously a market hole there.
GBA Detail Library
A collection of one thousand construction details organized by climate and house part
Search and download construction details
Replies
I've never looked at Panasonic. Buy I just noticed they do have 5k BTU wall head and slim duct units for a multi system.
Model # ?
CS-ME5RKUA is the wall head
Click catalog here:
https://na.panasonic.com/us/support/homeowner-support-resources
I'm also looking at Daikin now. They have some good 0.6 static slim duct units now with some decent looking -13f outdoor units in the Aurora line.
> way more energy efficient
Do you have data that shows that a 3K unit would be "way more efficient" than a 6K-12K unit modulating/cycling to produce the same outputs as a 3K?
The 6K Mitsubishi unit modulates down to only about 1.5k, as an single head, and probably can’t get that low as a multisplit. Now imagine you have a room with a ~2k cooling load on design day. on the median day, in many climate zones, you likely need <= 1k btu of cooling. So clearly a smaller unit would be better..
Also, you size outdoor units to indoor capacity, not load. So if the units are all 2x the size of there load, your outdoor unit is using 2x more electricity than you need.
So it only spends ~50% of the time on, which might be more efficient, put less hours/wear on the motor, and not create any noticeable temperature variation or reduction in latent removal.
While you may think they are "clearly" and "obviously" wrong, Chiltrix cycles their compressor even at outputs where they could operate continuously - because their data proves it is better.
> if the units are all 2x the size of there load, your outdoor unit is using 2x more electricity than you need.
For about 1/2 the time, so about the same kWh of electricity.
I don’t think your right here. These systems gain efficiency by being able to modulate. Having it constantly short cycling seems like it’s bad for everything. By your logic what’s wrong with 12k BTU or 24k BTU in every room then it’s 12x oversized for only 1/12th the time. Or 24x oversized for 1/24th the time. Clearly they felt a need to disambiguate between other sizes besides on the small side.
That is not anyone's logic
"Diminishing returns"
Certainly the most efficient system is one that is perfectly the size for a particular set of conditions, but as you get closer to 'perfect' the difference between the system you are using and a theoretical 'perfect' system gets smaller.
You seem to be forgetting that a compressor uses no power when it is off, the waste is primarily from the extra current used to start the compressor from off. A smaller compressor uses less energy to start so there is less waste in turning it on and off.
By your logic[cough cough strawman] any minisplit that cannot modulate down to 1 BTU is wasteful
>"I don’t think your right here. These systems gain efficiency by being able to modulate. Having it constantly short cycling seems like it’s bad for everything. "
A 50% duty cycle isn't short-cycling. As long as it's controlled to have some minimum on-cycle length the hit from the power use during idling/monitoring time between cycles is modest, but real.
Modulation below some level results in lower efficiency. According to the NEEP spreadsheet data LG's cold climate mini-splits are veryefficient at mid and max speed at +47F, but at their minimum modulation level of 1023 BTU/hr the COP drops below 1.5.
Comparing a Chiltrix to a mini-split isn't exactly apples to apples. Moving even small amounts of heat with blowers with air takes more energy than moving it with water. The Chiltrix only has the blower in the outdoor unit to manage. But the designers can (and apparently do) set the modulation limit at the low end to keep the efficiency from falling off a cliff when the loads are microscopic. From an efficiency point of view there are limits to modulation.
Ryan: dose can turn water into poison. Let's stick to the 2x being discussed.
I've seen (somewhere), data showing that 40% over-sizing of an inverter heat pump was beneficial. Clearly at some point things reverse, but nobody has any data (vs thought experiments) showing that up to 100% over-sizing (peak or rated output to design day load) of inverter heat pumps is an overall negative?
Those small ducted units would seem to show promise in high efficiency houses where single point heating proves difficult or inspectors balk at 'unheated' bedrooms
I can't reply to dana/keith, but, im not sure im intending to make the strawman argument, but, i guess i just have understood from reading here, that what matters is that the capacity distribution roughly meets the load distribution. If the min load while on is say, 250 BTU, median load is 750 BTU, and the max load is say 1.5k btu, and this follows a normal distribution, you don't want a unit whose minimum capacity is 1k btu, median is 3k and maximum is 6k. That seems like a real mismatch in sizes. The median excess BTU provided seems likely to be 60% of the load, and the min excess is like 50%.
It seems like the advice im to glean here is that, while this is a mismatch in load to capacity, and there is a perhaps noticeable performance hit, it's at the lowest end of the spectrum in terms of absolute size, so it's not that bad, I see what your saying, but, thats only try when you look at one room in isolation. it does make me wonder why they don't just make a product which can better match loads like this?
In particular, lets say these 2x mismatches are in most rooms of a home, means the "head in a room" solution lends itself to grossly oversized outdoor units. These efficiency losses seem like they will add up. think about it, if you only need 2 tons of cooling (as is the case in my upstairs area), but you end up installing 4 tons of capacity, that seems grossly oversized. It's exactly what most articles on GBA talk about when they talk about "right sizing" things? Or am i missing something.?
Mini-split performance at low load is an important question. I haven't seen much data on it from manufacturers. As far as I can tell, there are two effects at play.
1. Electric motor efficiencies tend to drop off precipitously at low load. See Figure 1 here: https://www.energy.gov/sites/prod/files/2014/04/f15/10097517.pdf or Figure 4 here: https://www.energy.gov/sites/prod/files/2014/04/f15/amo_motors_sourcebook_web.pdf.
2. Spinning up a motor/compressor causes spikes in current draw and mechanical stresses as the motor overcomes components' inertia. This temporarily increases electricity use and adds wear and tear. Some discussion here: https://www.energy.gov/sites/prod/files/2014/04/f15/motor_tip_sheet10.pdf
(Edit: I missed Dana's post upthread, where he mentions that standby energy use during 'off' times while cycling can be significant. I'll lump that in with #2.)
There's a trade-off between these two effects. #1 suggests cycling as a way to avoid running at low load/low efficiency. #2 suggests avoiding cycling to avoid power spikes, wear and tear, and (edit) standby energy use.
Different manufacturers seem to balance this trade-off differently. As Jon R mentioned, Chiltrix chooses to cycle even at moderate loads, presumably to avoid the steep drop-off in their motor's efficiency curve. As Dana Dorsett mentioned in another thread, LG does the opposite, modulating down to 1 kBTU/h even though their COP drops to 1.5 in that range, compared to 5+ at its load sweet spot. More discussion in the comments here: https://www.greenbuildingadvisor.com/question/visualizing-mini-split-performance-data
Regardless of the manufacturer's choices here, it seems that operating a heat pump at the low end of its load range should be avoided when possible. Size so load is in the middle/upper portion of the heat pump's range most of the time.
I'd be happy to change my mind about this, though! More data please :)
Thanks for sharing this! Essentially this affirms what i am saying, i think it's pretty obvious to me that a major manufacturer producing a 3k BTU head would fill a large hole in this minisplit market, it's surprising that they haven't done this. Someone with connections to mitsubishi from this community should get them to do this.
At that load range, you can get an EnergyStar window air conditioner for a couple hundred bucks and install it yourself. Maybe mini-split manufacturers don't think they can compete at that price point? Mini-split installation generally costs about a grand in labor alone. (That's not counting the heat pump itself or the installer's overhead and profit.)
Right, essentially failing to see the market. If it’s part of a larger system, also if it’s a window AC the air leakage will result in needing a larger system..
Recognizing you can't make any money with a certain product is not "failing to see the market". A market of one has no profit potential.
When you get down to whole house loads below 20k, not many people are going to opt to have a tiny head in every room. It's not necessary, it's ridiculously expensive and it's less efficient overall than having a couple of more modest size mini splits in general living areas. Or a ducted system, which will still be cheaper and efficient than multiple 3k heads. If your peak load is 1.5k, on the hottest and sunniest day of the year, do you even need cooling at all?
There's a floor to how cheap compressors, evaporators, coolant lines, condensers, etc. can get. At some point the energy you save vs just straight electric doesn't justify the initial cost. The consensus right now seems to be that 3kBTU/h is below that floor.
I'm with Ryan in that there's definitely a need and a market for smaller mini-split heat pumps if they can be priced right. As an apartment dweller and frequent traveler, I so don't want to put up with the noisy window-shaker HVACs that make me choose between sleep and some heat/cold. The window-shakers require a great big hole in the wall that conducts heat readily through it's metal frame and working parts to the outside. A mini-split needs only a very small hole that's easily insulated. The noisy compressor parts are outside. That's what I want in my hotel room or apt.
Right, what I meant was minisplits aren’t really in the same market as in wall ACs. There viewed by most as an improvement to central air. Many people retrofitting old homes without AC or adding them to new homes, want to ensure cooling in every room, the benefits over ducts is not needed to be repeated here. I was just saying that when I think about spending money on ductless, I’d be happy to spend the same price on a 3k BTU unit, of it will work better, and spend less money operating all the machines. Even if the COP at half load was a bit worse, you don’t pay for COP you pay for kilowatt hours, meaning you have some wiggle room, if the COP of the 3k BTU unit is above 1/2 COP of the 6k BTU unit on average, it’s still cheaper to run.
Kevin: most information (and both of your points) about non-inverter driven motors are very wrong when applied to inverter driven motors.
Don't underestimate marketing - a larger turn down ratio sounds better, even when it would be better to cycle (slowly).
"Kevin: most information (and both of your points) about non-inverter driven motors are very wrong when applied to inverter driven motors."
@Jon R, interesting. Do you have any data or references on that? Motors aren't my area of expertise, but I'm very interested in this question. I think it's an important point in real-world heat pump performance, since systems are so often oversized.
This DOE document,
https://www1.eere.energy.gov/manufacturing/tech_assistance/pdfs/motor_tip_sheet11.pdf,
says that efficiency drops off significantly at low load for electric motors driven by pulse-width-modulated VFDs. The 5 hp motor in Table 1, for example, is 35% efficient at 2% load, 80% efficient at 13% load, and over 90% efficient above 25% load.
But that's for pulse-width-modulated VFDs. Is that not relevant to the inverter-driven VFDs on most mini-splits? If not, could you point me to more relevant efficiency curves/data?
I'd research ECM or EC motors. Over reasonable ranges (say 20%-100% load) their efficiency curve is very flat. Startup is a complete non-issue.
Also, to your point upthread, Jon:
"I've seen (somewhere), data showing that 40% over-sizing of an inverter heat pump was beneficial. Clearly at some point things reverse, but nobody has any data (vs thought experiments) showing that up to 100% over-sizing (peak or rated output to design day load) of inverter heat pumps is an overall negative?"
The clearest real-world data I've seen on this from this 2015 DOE report, "Field Performance of Inverter-Driven Heat Pumps in Cold Climates". Link:
https://www1.eere.energy.gov/buildings/publications/pdfs/building_america/inverter-driven-heat-pumps-cold.pdf
In Section 4.2.5, Williamson and Aldrich wrote the following.
"Inverter-driven compressors can achieve high efficiencies because they have the ability to modulate output to match the heating load. However, when the load drops below the system’s minimum capacity, cycling can occur. Cycling is known to have detrimental effects on the efficiency of these units (Christensen et al. 2011).
At Site 2, a highly insulated Passive House that utilized the DHP as the only heating source, the DHP began to cycle when the outdoor temperature exceeded 40°F. This trend is clear in Figure 13 in the Results section; COP actually drops at higher temperatures when more cycling occurs. This home had a design load of 6 kBtu/h and a 1-ton system was installed (Mitsubishi FE12). Figure 42 displays the power consumption during one of these periods of modest loads (in April 2014). At these conditions (ODB = 55°F, indoor dry bulb = 72°F), the system’s capacity was limited at approximately 5,000 Btu/h. These operating conditions translated to a COP of 1.41 for the day."
I attached a copy of their Figure 13 for reference.
For what it's worth, I think undersizing mini-splits is also bad. Manufacturer data I've seen suggests that COPs drop off at very high load, too. A tiny mini-split that spends most of its time going full blast may perform worse than a huge one that spends most of its time cycling. Somewhere in between is a sweet spot.
So at near zero load on one model, COP with cycling dropped to 1.41 (from the graph in fig 13, I'd say such very low load COP = 1.5). But we also know that COP with modulation can drop below 1.5 (see #10 above). If I had to form an opinion from such limited data, I'd say "modulation to low levels vs even quite short cycling shows no difference in efficiency". Quite different from popular opinion.
Then there is the issue of looking at efficiency at single operating points - what really matters is seasonal efficiency (ie, even large efficiency differences don't matter if it's rarely operated at that point).
> Cycling is known to have detrimental effects on the efficiency of these units (Christensen et al. 2011).
But looking at that reference, I don't see any data regarding cycling.
One would think that some researcher would install two inverter mini-splits side-by-side, one 100% over-sized, and cause them to be active on alternate days. Measure energy draw, temperature fluctuation and humidity over a year.
> Somewhere in between is a sweet spot.
I agree. Probably pretty broad/flat (eg, little difference from 0% to +100% over-sizing) for inverter units.
"But looking at that reference, I don't see any data regarding cycling."
Judging from context, it seems that when the authors cite Christensen et al. 2011, "Field Monitoring Protocol: Mini-Split Heat Pumps", they're actually referring to Winkler 2011, "Fujitsu 12RLS and Mitsubishi FE12NA mini-split heat pumps". You're familiar with that one, I think. Jon Winkler was a co-author on the first report. The first one lays out an experiment protocol, the second has lab results. (Which, as you've pointed out elsewhere, are limited, at least as far as cycling goes.)
Correct me if I'm wrong, but I think we're pretty much in agreement on the big picture. Whether cycling or modulating, mini-split performance at very low load is probably bad. Ditto at very high load, with a range of sweet spots in between.
I'd be interested to see the results of the experiment you suggest.
I'll read about EC motors, thanks.
So you guys are saying you think that the COP of a 3k BTU ductless unit would be < 2 which is why they don't exist?
If a 3 kBTU/h mini-split operated full blast most of the time - because it served too big or drafty a space, say - I think its COP would suck. Does that have anything to do with why manufacturers don't make 'em? No idea!
Kevin, there is a load vs efficiency graph below. It suggests that avoiding 80+% load operation is more important than avoiding low load operation. No idea if residential heat pumps have similar curves.
https://www.csemag.com/articles/back-to-basics-vrf-systems
Thermal mass and thermostat dead-band/hysteresis will also play a role. Long cycles are efficient, even if the system is cycling with a duty cycle of 10%. Ie, substantial over-sizing can be perfectly fine - with enough thermal mass. But turn off AC for too long and the lack of moisture removal may become an issue.
If there is a problem with a 6K mini-split in a 3K room (we have no such data), perhaps the best solution is to increase thermal and moisture storage in the room.
Thanks for the link, Jon. That's a good point about thermal mass, cycle length and dehum.
"No idea if residential heat pumps have similar curves."
Mitsubishi mini-splits, at least, have efficiency curves that look similar at the high end to the one you linked. See the graph I attached.
That data is from the most recent 6, 9, 12 and 15 kBTU/h M-Series submittals and the NEEP spreadsheet. Note that the horizontal axes are percent of *rated* capacity, rather than of maximum capacity. Maximum capacities are quite a bit higher than rated. Also note that the axis limits are a bit different for heating vs. cooling.
I haven't been able to find any data published by Mitsubishi on COP below ~20% of rated capacity (when Mitsubishi mini-splits start cycling), anywhere. The dashed lines at the low ends of those graphs are just a best guess based on (sparse) laboratory tests and field experience reported in the research literature.
>"I haven't been able to find any data published by Mitsubishi on COP below ~20% of rated capacity "
Not looking hard enough at the NEEP spreadsheet, mayhaps?
At +47F the FH09 can operate down to 1,600 BTU/hr which is more than 20% of it's rated capacity of 10,900 BTU/hr.
Per the NEEP data at +47F the COP efficiency at 1,600 BTU/hr is 4.26, which is below the COP of 4.50 @ 10,200 BTU/hr, but above the COP 3.58 at it's maximum of 18,000 BTU/hr.
At +17F it can operate below 20% of "rated too. It will throttle down to 920 BTU/hr with a COP of 3.37, which is above the COP 3.27 at the "rated" 6,700 BTU/hr, and well above the COP 2.48 at it's maximum 12,200 BTU/hr. T
At +5F there is no "rated", but at mininum output of 640 BTU/hr (the heat output of two standing adult humans) it's COP efficiency is 3.13, and at it's max of 10,900 BTU/hr it's only 1.35.
To see some of the limitations of modulation, look at the NEEP spreadsheet for LG/s LA090HSV5 ( LAN090HSV5 + LAU090HSV5 ) at +47F. At it's minimum 1023 BTU/hr it's only pulling a COP of 1.5, compared to a COP of 5.60 at it's rated 10,900 bTU/hr.
LG's LAN/ LAU090HYV1 seems to stay above COP 4 though, from it's 1,023 BTU/hr min to it's 20,472 BTU/hr max, which is probably due to the vapor injection compressor technology used and how it is controlled.
Let's say one arbitrarily decided that seasonal performance = 0.01A+0.42B+0.45C+0.12D
Where:
A = COP @ 100% Load
B = COP @ 75% Load
C = COP @ 50% Load
D = COP @ 25% Load
Attempting to use rough A,B,C,D values from the heating graph that Kevin posted:
Undersized - 67% (150%, 112.5%, 75%, 37.5%): 82, 97, 100, 98
Exact size - 100% (rated capacity = design load): 98, 100, 99, 97
Oversized - 200% (50%, 37.5%, 25%, 12.5%): 99, 98, 95, 90
I get seasonal performance of:
Undersized - 67%: 97.6%
Exact size - 100%: 98.3%
Oversized - 200%: 94.8%
My conclusion:
All similar. WRT efficiency, it makes much more sense to worry about head mounting height than exact matching of HP rating to load.
"Not looking hard enough at the NEEP spreadsheet, mayhaps?"
Mayhaps, Dana! I believe, though, that I used all the NEEP data available for the the 6 - 15 kBTU/h units (hence the data points in the 15-20% range on the heating graph). I should clarify that what I'm really interested in isn't COP at minimum modulating capacity, but COP while cycling.
I agree that the LG case is a good example of the potential efficiency hit from modulating down to very low capacity.
>"I agree that the LG case is a good example of the potential efficiency hit from modulating down to very low capacity."
But the OTHER LG case (the LAN/ LAU090HYV1 ) is an example of why you can't model vapor injection compressors in the same way as simpler 1 & 2 stage compressors, since there are multiple axis' at play in the compressor control, not just compressor speed. That's why the Mitsubishi H2i and Fujitsus don't take a big hit at minimum speed too, but the designs and control algorithms are proprietary- making a generic model that fits all products is difficult or impossible.
Jon, those number seem plausible to me.
Another potential upside of oversizing is that it's more conducive to thermostat setbacks. An undersized unit would have to work at high capacity/low COP to recover from setbacks, eroding some or all of the potential savings. I believe one of the field monitoring reports showed lower seasonal-average COPs for mini-splits in homes that used night setbacks. That might not be an issue, or as big of an issue, for an oversized heat pump.
It's also worth noting that bigger heat pumps cost more up front. COP is important, but of course capital cost is too.
Edit: here's the study I was thinking of.
https://www1.eere.energy.gov/buildings/publications/pdfs/building_america/monitoring-mini-split-ductless-heatpumps.pdf
"Temperature setbacks (on/off operation): Previous work has shown that deep temperature setbacks of simplified heating systems can exacerbate temperature unevenness issues. One homeowner complained of temperature unevenness; when the data were examined it was clear that they operated their MSHP in an “on-off” manner rather than using a fixed set point. This approach resulted in wide swings in interior temperature (between 60°F and 70°F+). The electricity use showed many hours with the MSHP running at maximum capacity followed by periods with the unit shut off. When operated in this manner, the MSHP is heating at its least efficient (maximum output) state. Electricity consumption was a high consumption outlier; when compared with simulations it was the worst-performing house—the heating use was 57% higher than simulation."
Similar effects were reported in http://vermontfuel.com/heatpump/ewExternalFiles/DOE_inverter-driven-heat-pumps-cold_2015.pdf (see section 4.2.3).
I'm not sure anyone knows yet whether thermostat setbacks with mini-splits are an overall win, in terms of cost or carbon emissions.
These are very interesting discussions. I was going back and forth between the 9k and 12k Fujitsu slim duct. They both go down to 3.1k, buy there isn't a vast diff in the max cooling output. Rated is of course 3k more.
Now that I know energy efficiency seems to follow rated and lower capacity, it seems the 12k system is worth the extra measely $100 for the equipment when the coolled area is already at the 9k limit or just past when derated for line length.
Extra dehumidifiation capacity to boot. Coming up to speed after setback is a great point too.
> whether thermostat setbacks with mini-splits are an overall win
Agreed, it's more complicated than a simple "it will cause the mini-split to run at full capacity and increase net energy use". For example, imagine the case where one could set back 5F at night, leave it set back while while getting ready for work and then slowly set it up in the afternoon (when it's warmer outside). Lots of hours at lower temp, higher COP, no full power operation. Probably with a net increase in comfort.
Sorry to bring back an old question, but, I have an answer from a mitsubishi application engineer. He said that the US residential units are sized generally for heating, and that generally heating loads are at least 2x cooling loads in places that need heating, or on the cooling side it’s so hot out that smaller units are rarely needed. He also doesnt think their is a market for the smaller units.
>"... generally heating loads are at least 2x cooling loads in places that need heating."
BS. Did he point to any real world statistics on that?
Reads like a "marketing engineering" statement designed to make you go away. Very few bedrooms really need a 6000 BTU/hr (or larger) head for either cooling or heating.
>"He also doesnt think their is a market for the smaller units."
That part is probably true- most people have the "bigger is better" mentality, and are used to cooling their bedroom with a half-ton window-shaker, even though a modulating unit half that size might deliver better comfort.