Cooling Load Calc for Sizing a Ductless Minisplit
Last winter, I received a heating & cooling load calc for my house. For cooling, the design temp was set to 93F and the indoor temp to 72F, however the ASHRAE design temp for the Detroit area is actually 87F, and I’ve read that indoor temp for cooling load calc should be set to 75F. In other words, the HVAC guy was erring on the side of oversizing, which apparently is very common.
However, I’ve also read that for an inverter-driven system such as ductless minisplits, it’s OK to oversize a bit because the system can run lower than the rated capacity, and in fact is more efficient when running closer to 50% (?) of rated. (Obviously, it can be too oversized if the minimum capacity exceeds the load.)
With those two things in mind, I’m wondering… should this ductless system design be based on the actual 1% temps for Detroit, or the more extreme temps (0.1%?)? I don’t know which would be better in terms of both comfort and efficiency. Do the ductless system ratings already take design temps into account? For example, I noticed that Mitsubishi’s 9K unit is rated for 9, but has a max capacity of 12K. This also happens to correspond with my upstairs load calcs… roughly 9K at ASHRAE design temps, and 12K using more extreme temps. Does that mean the 9K model would have peak performance, yet still be able to handle the extreme days? Or would it be better to use the 12K model and not have the 9K’s inverter hitting max capacity (i.e. lower efficiency) on the high temp days?
GBA Detail Library
A collection of one thousand construction details organized by climate and house part
Replies
The "rated" 9000 BTU/hr is the modulation rate at which the efficiency was tested at the 80F indoors, 95F outdoors SEER test conditions. The 12,000 BTU/hr is it's maximum performance under those same conditions.
That's a temperature difference of 15F. If the indoor temp is 75F it won't put out quite as much at 95F outdoors, but it'll put out about that much at 90F (75F + 15F =90F.)
The modulation range, particularly the MINIMUM modulated output is at least as important as the max number, since that determines when it stops modulating, and starts to cycle on/off, which takes a toll on as-used efficiency. If you bump up to the 1-ton Mitsubishi it is likely to have a higher minimum modulation, which means you'll lose some efficiency on low-load days. If those low-load days also happen to be high humidity, the cycling means it's not drying continuously, which is lower comfort as well.
The breakdown of latent & sensible cooling needs to be understood to know for sure, but it's likely that the 3/4 ton unit is more ideally sized to your loads.
Any high SEER cooling equipment can have problems keeping up with the latent loads (humidity) in high-humidity climates. Detroit is in the middle, but definitely has it's share of high-humidity high-cooling load days. Ductless mini-splits usually have a "DRY" or "DEHUMIDFY" cooling mode, and it's likely that you'll have to use that mode to stay comfortable on the stickiest days. The efficiency isn't as high when operated in that mode, but if you right-size rather than oversize, it'll usually keep up with the latent loads without resorting to a lower efficiency mode.
What is a good way to determine what the lower loads might be? Would the load calc software used by the HVAC guy likely be able to determine that? Should I ask to see the numbers with a design temp of 81F, for example? That's roughly around when we might turn on AC.
For reference, the 93F load totals I'm looking at are: 10.6K upstairs, 15K downstairs, and 3.5K latent (not specified per room as far as I can tell). I was just splitting the latent 50/50 for upstairs & downstairs for estimate purposes.
I was just estimating the 87F load by proportionally reducing the sensible load by the delta T difference (i.e. 15 dT / 21 dT = 71%). Not sure how accurate that is.
Speaking of minimum modulation, do you happen to know how that works in the case of multi-zone units? For example, if an FH09 indoor unit (1.5K min) was attached to a 3C24 HZ unit (6K min), could it still potentially output as low as 1.5K, assuming that other heads were calling for cold air and using some of that 6K?
Thanks again for all of your help, Dana.
If the software used by the HVAC guy is based on Manual-J (most are), and the relevant data are entered correctly, yes, you can use that package to come up with reasonable load numbers at different indoor & outdoor conditions.
With cooling load direct solar gains through windows are a large fraction of the total (often the lion's share), and solar gain doesn't change much with delta-T, so scaling it by the delta-T isn't very accurate- it'll be more than 71%.
That said, if the calculated load at 93F outside/72F inside was 10.6K, you don't need anything bigger than the 3/4 tonner to handle 87F outside/75F inside, since it has the capacity for even the much bigger delta-T. The load at the bigger delta-T is only 18% bigger than the rated-efficiency-tested modulation level, and 13% lower than the maximum output level of that unit. So even if the load at 87F out/75F in turns out to be only 15% lower rather than the calculated load (only half the presumed reduction using simple scaling), it would still be operating near it's efficiency-tested modulation level at your design condition, and have sufficient margin to cover the shared latent load & higher temperature hours.
If it's the FH09 you're looking at, that's the right size for that zone, not the FH12 (though it wouldn't be too terrible), and not the FH06, which won't quite keep up under high-temp + high-humidity conditions, even if it just squeaks in (or close to it) at design condition.
On a multi-split if only the FH06 head is calling for cooling, it will be forced to take the whole 6K output of the compressor. That makes the FH06 run at it's max speed to keep from short-cycling the compressor, and the head & compressor cycle on/off together, but the compressor will be running at it's highest efficiency, while the FH06 would be at it's lowest. It's not ideal, since they can't modulate with load,and efficiency will suffer. If other heads on the system are calling for enough cooling to cover 4.5K of the compressor output, the FH06 could modulate to it's lower level. With multi-splits the minimum modulated outputs of the compressor can be higher than the whole-house load much of the time, but with three heads it'll be only slightly less efficient than three separate mini-splits, unless you start turning one or two mini-splits completely off when the loads are low.
Another issue that comes up with multi-splits is that it's sometimes difficult to keep the refrigerant line lengths below the specified maximums, which reduces the maximum capacity of the head, and it takes a hit in efficiency. With separate mini-splits there can be more flexibility for siting the outdoor units closer to their indoor heads.
I got tired of waiting for the HVAC contractor to give me the load calcs with the correct Detroit temps... so I went and bought a $50 2-month license for HVAC-Calc, which is supposed to be Manual J. It looks like the program is in the ballpark, based on calculated vs. actual heating load that I recorded on an 8F day (boiler output was 34K, my calc was 38K but that doesn't account for ~3K of people and appliances).
So here is my calculated load at Detroit's design temp of 87F / 75F indoor. I included 5 people downstairs (4 bedroom house + 1), kitchen appliance 1200 BTU/h, and 800 BTU/h plasma TV (I've watched it heat the living room on multiple occasions). I also assumed all window shades were open.
First Floor: 12355 BTU/h
Second Floor: 6651 BTU/h
Basement: 308 (negligible)
Of course, if everyone were upstairs during the hottest part of the day (unlikely), around 2K of that load would shift from First to Second. I also played with the temps, to compare to the HVAC guy's calcs... using his 93/72F temps adds about 2K to First Floor and 1.4K to Second Floor.
The minimum caculated loads, after lowering the outdoor temp to 81 and having shades closed upstairs during the day, with no appliances on / people around, gives me these numbers:
First Floor: 6448
Second Floor: 3677
So it appears that any decently sized mini-split is going to handle the minimum loads fine, without much short-cycling. Based on these numbers, I am thinking of a 12K unit for the First Floor, and a 9K (or 6K?!) for the Second Floor. I'm pretty certain the 12K would work well down there, because on an 87F day, with everyone (+ guest) home and downstairs, and cooking dinner, and TV on... the load is only 12355. It's only if it happens to be 93F outside, the thermostat is set to 72F, AND all the rest, then the load (14382) exceeds the max output.
I'm not sure about the Second Floor though... design day temps put the load at 6651... but once you start throwing people, TVs, etc up there, along with higher temps, it could be more like 10K. Maybe the 9K unit would be a safer bet, on the off-chance that the downstairs 12K unit is under huge strain to keep up (ex. hosting for holiday), and some of that excess heat is escaping upstairs.
Any suggestions for sizing the mini-splits based on my numbers? Or any suggestions for numbers to plug into the software to take a look at different scenarios?
Thanks!
having lived with minisplits for AC and occasional heat for 10+ years I would suggest no over sizing. The units struggle to modulate well when oversized. I believe the manual J is considered over sized by nature, so further oversizing is even worse.
The smallest they made in 2009 was 9k and the one in my largish master bedroom is run with the fan on low and the vanes kicked to the side to avoid frostbite to the beds inhabitants
The hidden mini splits have become much more common and more efficient since and might be a good answer
Hi Keith, thanks for the response. I've read that Manual J has a bit of safety built into it, so the calculated loads might be a little on the large side. Also, our house gets a lot of shade from the surrounding trees, so the sun only shines directly into certain windows for brief periods (1-2 hours) throughout the day. So I think you're right that the numbers I'm coming up with are on the larger side--even though it's less than half of what some HVAC contractors were coming up with!
As far as modulating, both the 6K and the 9K units from Mitsubishi would be able to go as low as 1700 BTU/h, though Fujitsu's 9K has a minimum of 3100.
What I'm looking to do is install a mini-split on each level of our colonial (minus basement). We're about to redo the kitchen to make it open concept, which should make almost the whole floor accessible by a single mini-split. However, the upstairs is a tougher nut to crack, because of the separation of each bedroom, potentially closed doors, etc. I looked into the hidden/ducted systems, but they start at 9K output, which is about as large as I'd want to go. Also, this is a retrofit, and our upstairs ceiling is only 7.5', so it's a pain coming up with a duct design that doesn't impede on living space. Most likely we'll install a unit at the end of the central upstairs hallway, and keep bedroom doors open during the day for cooling. If the load is actually 4-6K, then the cool hallway idea should work (I think...).
The time of day of the peak load affects the likely occupancy levels. Typical peaks occur between 1-3PM- is that really when people are going to be in their bedrooms running the video game console?
What is the heat load of the upstairs zone? There isn't much cost difference between a FH06 and an FH09- about $100. If the FH09 pretty much covers the heat load and the FH06 doesn't it's $100 well spent.
Some of the first floor zone's load may be covered by the upstairs mini-split from convection, assuming that the stairwell is open. You'll do just fine with the FH09 on the first floor too. But the FH12 isn't extreme overkill, and would still have you covered even on those hot summer afternoons when you have a dozen 200lb Czechs over for a polka party. :-)
You kid... but that scenario might not be far off haha. Then again, we could just as easily make the basement into the gaming area for those summer days when it's too hot to go outside.
I was assuming that peak times were like 3-6PM, since the outside of the house has been soaking up heat all day, and the sun is a bit lower in the sky so it shines directly into windows more easily. But if I recall, we get solar gain in different rooms at different times (ex. 1-3 south-facing rooms downstairs, 5-7 west-facing rooms upstairs).
Maybe I should try looking at the numbers with different likely scenarios, to see what "luxuries" we would be willing to give up. For example, I doubt we need the ability to cram everyone on the same floor, with several TVs and computers running, blinds open, all while it's 93F outside. At the very least we could close some blinds if it gets too hot. Oh, and probably don't need to host polka parties anytime soon :P
FYI the heat load at 4F design temp:
First Floor: 23821
Second Floor: 11223
Basement: 6463
At 40F design temp (when temps drop below this, likely to use boiler):
First Floor: 10827
Second Floor: 5103
Basement: 2939
I wonder if I could get away with 2x Fujitsu 9RLS3 units. Though they don't modulate as low as the FH09s, it looks like the lowest I can get my loads down to is 3.5-4K per floor (no occupancy, no appliances, shades drawn), which is under the 3K minimum. Though maybe the 1.7K minimum would get used more often, for example, if we shut the door to the spare bedroom to avoid heating/cooling it.
One advantage to the Fujitsus might be cost... I've heard they are generally cheaper to install, plus I wouldn't need the Hyper Heating feature on the FHs.
Host the polka parties at 5AM on the coldest day of the year, then! :-)
With a design heat load of ~11K @ +4F spending another Franklin for the 10.9K @ +5F max capacity of the FH09 vs. 8.7K @ +5F for the FH06 is probably going to be "worth it" for the upstairs zone.
And with a heat load of 23K @ +4F for the first floor you'll definitely want the 1-ton for the first floor too (Czechs or no Czechs.)
If it was going to be your primary heat source even the FH18 (20.3K @ +5F max) would make sense for the first floor, but that's a bigger upcharge. The FH18 a bit overkill for the cooling load though, and would cycle quite a bit during the cooling season.
Is the FH06 any better than the FH09 when it comes to efficiency, or does it just look better on paper because it was rated at a lower capacity?
Are the Fujitsu units any better in terms of efficiency, and for heating? It looks like there are plenty of Mitsubishi and Fujitsu dealers in my area, so I'm guessing neither should be a problem for service. We even have an "Elite Contractor" for Fujitsu in my area!
I definitely like the idea of using the heat pumps for the shoulder season... but I'm not entirely sold on using it in the middle of winter. The main reason is that with the upstairs being served by a single unit in the hall, I'm worried that bedrooms with closed doors will end up getting too chilly at night (when temps are lowest). That isn't as big of an issue during the cooling season, since the load should become lighter after dark. Also, it may not matter as much in the shoulder season since loads will generally be small anyway.
When we upgraded to a high efficiency boiler, we also had them separate the system into two zones, which will match the two mini-split zones. So I guess if we wanted, we could also play around with some combination system. For example, if closed bedroom doors was an issue at night, we could just heat the upstairs with the boiler instead.
I suspect it's the exactly same coils & blowers in both the indoor & outdoor units for both the FH06 & FH09, possibly a smaller compressor and variable refrigerant valve for the FH06. Oversized coils combined with a lower operating speed & smaller compressor will give the FH06 an edge in the bench test efficiency.
Fujitsus only throttle back to 3100 BTU/hr, which means the 3/4 tonner will modulate less than the Mitsubishis. The coils of the 9RLS3 are probably the same as the 12RLS3, and a bit bigger than the FH09 (all of this is speculation based on physical size and specified capacities.) If you had enough load to keep it modulating the Fujitsu might beat the Mitsubishi on efficiency, which would be the case for the first floor cooling loads, but maybe not upstairs. For heating the 1 ton Fujitsu downstairs and a 3/4 ton upstairs would probably work fine, but in cooling mode the Fujitsu will cycle a bit more.
The 12RLS3 has substantially more heating capacity at +17F than the FH12NA (16K vs. 13.6K), and if the upstairs mini-split is carrying a substantial fraction of the first floor's cooling load via convection up an open stairwell, the higher minimum-modulation won't really be much of an issue.
If you're planning to run them many hours when it's below 20F outside be sure to specify the -H versions of the Fujitsu (indicating that it comes with a base pan defrost heater), and make sure that the Mitsubishi have MAC-640BH-U base pan heaters installed. Pan heaters kick on during defrost cycles to ensure that ice doesn't build up to fan or coil damaging levels during extensive cold weather use.