Using the “rated” or “max/maximum” BTU capacity when sizing a mini-split
michaelbluejay
| Posted in Green Products and Materials on
Here’s what I’m including in my article on HVAC sizing. I’m submitting it here for peer-review. Thoughts?
——————-
Confusingly, equipment specs show both a “rated” BTU capacity as well as a “maximum” capacity. For example, a mini split I bought has a “rated” cooling capacity of 6000 BTU, and a “max” capacity of 11,184 BTU. Which figure do you use when matching the equipment to your Manual J?
The short answer is that you can use either, but in general I feel it’s better to match for max. [see details in the article linked below]
GBA Detail Library
A collection of one thousand construction details organized by climate and house part
Search and download construction details
Replies
Deleted
So, after posting the above, I rewrote huge chunks of the article. I think there's more practical detail in my HVAC-sizing article than any other article on the same topic that I've seen. At the same time, I'm not infallible, so I hope some of you will check it out and share your thoughts, if I'm able to share the link and not have it rejected as spam: https://michaelbluejay.com/electricity/hvac-sizing.html
ManJ is very conservative, right sized equipment is already oversized, so using rated capacity will mean even further oversizing. For any sizing always use the max number.
The other thing is that chart. Except for a uninsulated pole barn, there is no livable house in cold climate that will have 40-60BTU/sqft heat loss. I've done fuel use based heat calcs on somewhat air sealed uninsulated brick house in zone 5 and was 17btu/sqft. Some for the rest there:
https://www.energyvanguard.com/blog/air-conditioner-sizing-rules-of-thumb-must-die/
Deleted
Deleted
Deleted
Deleted
I also looked at your chart and question its validity. My house, which is basically in the blue zone (though I'm in Canada), and its heating load is 5BTU/hr/sqft, which is half of the bottom end of the range listed. For cooling, according to the chart I need 4 tons. My house actually does just fine with a single 3/4 ton unit. There isn't even a range on the chart for cooling, it's just a single load number for any given square footage.
I think you should just delete method 2 altogether. In order to account for all the different variables, you'd end up with ranges so wide that they aren't even useful.
When I first read it, I missed the part about positing it for review. It seemed to be asking a question then posting a reply with a link to your site.
My statement about the accuracy of the article still stand. Not a good way to size.
Getting the sizing right requires work, no way around it.
I'll remove my earlier comment.
"My statement about the accuracy of the article still stand. Not a good way to size."
Then it seems like you didn't actually read the article (or even the title of the table).
What you guys are referring to my "chart", I think you actually mean the third (of three) *tables*. There are no charts in the article.
With a rough heat load calculation, I can easily get 40 BTU/sf for an old house in CZ-5 (not build to proper standards):
- Two-story house, 1200sf each, 2400sf total
- R-9.3 walls (2x4 w/R-11 insulation)
- R-19 attic
- 126sf windows @ R-1
- 4 doors @ R-4
- Uninsulated basement, 4' above grade, 4' below grade
- 3 ACH
- Ducts in attic, 10% loss
Of the 99k BTU, 36k is air infiltration and 24k is the above-grade portion of the basement.
Of course it's possible that I made a calculation error.
I like maths, lets calculate.
CZ5 so say 5F design temp so 65F delta. I doubt you'll find too many R1 windows out there, all these old houses have a storms, so at least R2.
2400sqft generally means about that many sqft of walls.
-2400sqft - 126sqft windows=2200sqft * 65f/R9.3=15000BTU
-126sqft of windows so 126*65/2=4100BTU
-uninsulated basement is hard to get exact number, lot depends how much above grade and how many layers of stone/brick, but say about 10000BTU loss.
-old house won't be 3ACH, closer to 8ACH or 2500CFM @ 50PA, cold climate N factor for two story is about 12, so 2500/12=200 CFM. 200CFM*1.08*65F=14000BTU.
-ceiling 1200*65/R19=4100BTU
Total is 47KBTU or 18btu/sqft.
I'm in Zone 5. Earlier today I was walking by an older rowhouse (so neighbor on both sides) that got a new heat pump installed. These are pretty small, generally 1000-1500sqft. I have done heat calcs for a similar place and the heat loss is around 1.5ton. The unit was a 5 ton hyper heat in the front yard, I think the installer might have been reading your blog.
Re: Basement, I did specify 4' above and 4' below grade. Martin suggests U-0.06 below-grade and U-0.67 above-grade. ("How to Perform a Heat-Loss Calculation, Part 2"). With a 34.65' x 34.65' footprint, that's 554sf of wall above and below. ∆65°F x 554 x (0.06 & 0.67) = 2161 & 24,127 BTU respectively.
Re: Air exfiltration, I'm using the formula Q = [∆°F] x [ACH] x [volume] x 0.018 [constant of the heat capacity of air], which I've found in various sources, including GBA. ∆65°F x 3 ACH x (34.65'^2 x 16' high) x 0.018 = 67k BTU. Perhaps I've made an error here, I'm happy to be corrected if I have.
"I think the installer might have been reading your blog."
First of all, it's not a blog. It's as much of a "blog" as GBA. Second, and I'm repeating myself here, but I know who's *not* reading my website (you), since what you keep ascribing to me is the polar opposite of what I actually say. This is like the sixteenth time you've completely missed what I said or what you saw:
1: That I asked for peer review.
2: That I specified the above- and below-grade portions of te basement.
3: That my site is obviously not a blog.
4-14: That my article says no less than ELEVEN times (including in the title of the third table, which you call a chart), that either Manual J is the gold standard or other methods are poor, or both.)
15-16: I pointed out in comment #10 above, twice, that my article doesn't say what you think it says, yet you apparently missed that...again.
Regardless of what else you say in the article, or how many times, you've got the table as a method for selecting equipment, and it's clearly biased towards oversized. You describe it as a poor method (which I think was a later edit, but can't say for sure), but why is it even there? And if it's there, shouldn't it at least be the best representation of the concept? You provided an example justifying the worst case in the blue zone. I'd like to see an example for the red zone. What kind of structure in southern Texas needs 37BTU/h/sqft? A camping tent? It doesn't even make sense. The difference in heating requirements between similar houses in Texas and Minnesota is only 10%? A house in Florida only needs 20% more cooling than one in North Dakota? And all houses of the same size in the same zone require the exact same cooling? All of these are unavoidable conclusions based on the table. I just don't understand the point of its conclusion. Even though you say it's a poor method, people are still going to look at it and decide it's better than nothing. But it's not better that nothing, it's possibly worse than nothing.
Like I said, basement is hard. Older houses typically have 3 or 4 wythe brick or two foot thick stone foundation, so usually around R3/R4 or so. The place is also not actively conditioned so a bit lower delta T. Maybe it can be 24kBTU for 1200sqft, but most likely much less. At least definitely less based on fuel use numbers and heat loss in the rest of the house.
Air leakage is measured at pressure (50PA). For heat loss you need to your natural air leakage or AchNat. There isn't a right way to convert between the two but you can get into the ballpark:
https://www.greenbuildingadvisor.com/question/what-is-n-factor
Since the actual air infiltration is a fraction of the ACH50 number, the heat loss is much less.
You'll have a hard time to come up with a house with real world energy intensity much above 25BTU/sqf even in cold climate.
As for the original question. Size based on rated or max there is no answer without actually running temperature bins and figure out how much time is spent at what COP and how much time is spent low load cycling. This can be done but not easy. Generally the oversized unit will spend more time cycling at low load which will cost more energy then any COP improvement.
For example on a project a while back, going from 0.95x size(right sized unit when combined with bath floor heat) to 1.4x oversize (next size up with nameplate capacity to match design loss), increased the low load cycling from 30% to 58%. What that translates to in real world energy use, is hard to say but definitely not a good thing and definitely a decrease in comfort.
Of course this assumes ASHP. Any air to water heat pump will have a definite improvement in COP from oversizing as long as the buffer tank is sized correctly and properly set up and managed.
PS. Sometimes equipment next size up will have similar min capacity, in this case there is usually an improvement of overall COP without increasing low load cycling, in that case by all means size up.
It's calculable.
If you go to NEEP.org and use their advanced sizing tool, where you pick a heat pump and put in your zip code and heating load, it spits out a graph that shows BTU/yr at each temperature.
The site also gives COP at min and max capacities for a range of temperatures. From that you can estimate your COP at each temperature. COP tends to be pretty linear, as does capacity. So you can find unit's min and max capacity at that temperature by extrapolating from the known datapoints. Then find COP at min and max capacity by extrapolating those too. Find your heating load at that temperature, and extrapolate between the COP at min and max for that temperature. Then multiply COP by annual heating load at that temperature. Do that for every temperature and add them all up.
The only thing that makes this hard is that NEEP doesn't seem to have a way to download the annual heating load data, you'd have to mouse over each point and write down the numbers. Other than that it could all be done in a spreadsheet.
The NEEP tool has been updated and now you can download the data in a CSV or XLS files. The only extra work is to map COP to each temperature bin and load. Theoretically, except for low load cycling, you could get yearly power use with a bit of ExcelFoo.
Indeed it does. I hadn't noticed that.
I think I've shared this before, but this is a spreadsheet that does that calculation. I just assumed temperatures were normally distributed and used the 99th percentile temperature and annual average to create a distribution, this data would be more accurate. Link is:
https://docs.google.com/spreadsheets/d/1sIpGCu2Rkb-SqlzY0vw-DUIAQGv8KMxMXG3p1QjuAaQ/edit?usp=sharing
------------------
AIR EXFILTRATION
------------------
Thank you for the explanation about air exfiltration. I now understand the difference between ACH50 and ACH-natural, and how to do a rough conversion. From that, I'm able to get the BTU from the sample 2400sf home down to 8k BTU.
------------------
FLOOR LOSSES
------------------
Let's try to make this productive. I'm happy to be corrected if my calculations are off.
Let's take a home in the warmest climate zone, which includes S. Texas as suggested. Design is 30°F/70°F, so ∆40°F.
House footprint is 1200sf with a pier & beam foundation, 1" stucco skirt, and 3/4" hardwood flooring. I own two houses of similar size in the same climate zone with the same foundation and flooring.
ArchToolbox shows R-0.68 for 3/4 hardwood and R-0.26 for 1" concrete. That's R-0.94 combined for losses through the flooring and skirt walls.
So losses through floor looks to be: Q = 1200sf x ∆40°F ÷ R-0.94 = 51k BTU/hr.
For a 1200sf house, that's 42.5 BTU/sf *just for the floor*, and that's before accounting for the vents in the skirting, or air leaks through the hardwoods.
But I agree that that seems abnormally high. What am I missing?
Normally either the floor or the walls of the crawlspace would be insulated.
Houses with uninsulated floors and crawls are COMMON here (like I said, I own two such houses), and I'm modeling for both an "efficient" and "inefficient" house, so the inefficient house has an uninsulated crawl.
Are my calcs correct? If so, then why am I being given a hard time by saying that that result is impossible, when it's not? (e.g., "You'll have a hard time to come up with a house with real world energy intensity much above 25BTU/sqf even in cold climate.")
Not hard at all, quite easy to even DOUBLE that, for old home construction that's common here, if my math is right. And if my math is wrong, then I'm all ears.
I assume the skirt goes around the crawl space? If that's the case you can't just add the r-values because the area of the skirt is not the same as the area of the floor.
What you have to do is calculate the heat loss through the floor and through the skirt and figure out what temperature they balance at.
Let's say the house is 30x40 and the skirt is 2' high, so the area of the skirt is 280 square feet. If the crawl space is at temperature T, the flow from the house into the crawl space through the floor is:
(70-T)* 1200/0.68
And the flow from the crawl space to outdoors through the skirt is:
(T-30)*280/0.26.
Setting those two equal and solving for T I get T=54.8F and the flow is 26.7k BTU/hr.
This a good resource for assembly R values:
https://www.ekotrope.com/r-value-calculator/
An uninsulated floor is about R3, uninsulated skirt is around R1.5.
Those types of floors are common in cottage country around for 3 season places. They are pretty much non-existent in anything four season as it is not livable, they always get insulated.
There is no such thing as R.94, for the simple fact you are ignoring the air layer on both sides of the floor.
interior floor something like .9
exterior of the floor subject to wind something like .17
A single pane of glass has an R value of virtually zero. It is the air layers on either side that give it a value of about 1
We get sloppy with saying things like an 'r13 wall' when it is a lot more complex than that
Please do more research before you get all wound up
And when R-values are that low small differences in assumptions make a huge difference in the outcome. Is your floor really exactly 0.75 inches thick? Or is it 0.8?
This article gives a method for calculating actual heat loss based on fuel usage:
https://www.greenbuildingadvisor.com/article/replacing-a-furnace-or-boiler
In the article, you list methods of sizing equipment. I would add the method above with the rating "good."
You also say, "Size it to your existing system (good)." I'd categorize it as mediocre. Yes, this is what most professionals mostly do. And yes, it should be no worse than the existing equipment. The problem is that most people don't really pay attention to their equipment unless it's not working. So if it's currently oversized they're not going to fix that.
Good point, thank you. I took your suggestion and labeled that method as "questionable", noting that anyone's existing system is likely already oversized.
Using existing furnace to size is a really bad idea. For example, my brother in law recently replaced a 70k furnace with a 2 ton heat pump. The cost delta of a 6ton heat pump would have been quite significant never mind all the issues with the significant oversizing. Usually the fuel burner oversizing is not that large but 2x is pretty common if not the norm.
The best way to size replacement units is by doing the fuel use calculation. It is the closest you can get to proper sizing as it eliminates all the building assembly unknowns. It just works.
Next best thing is thermostat runtimes of the existing fuel burner. With that you still have to compensate for weather when run and average of a bunch of cold days.
Thank you for the floor/skirt calculation DCcontrarian, it's helpful, and I follow the math. Of course now I really need to account for losses through the crawlspace vents, but I can't find a great source for crawlspace ACH or for crawlspace temperature vs. outside. I also couldn't find how Manual J figures vented crawls. The best I could find was this article which said that much of the info came from Allison Bailes' book:
https://hvacrschool.com/manual-j-field-data-floors/
The illustration in that article shows 40°F outside and 25°F crawl, for ∆15°F.
So for the 1200sf floor, losses would be (70°F - 30° - 15°F) x 1200sf ÷ R-0.68 = 44k BTU.
Adding up the other loads in the 2400 (2-story) house, that puts the house at 32 BTU per square foot. I re-ran the numbers for all climate zones and adjusted the upper range figures downward accordingly. I know they might not still be as low as some would like, but I'm happy to be corrected if someone can show that my math is wrong. Here are the details I'm using for the inefficient house:
• Winter design: 6, 12, 14, 25, 30°F (Zones 1-5)
• Winter set: 70
• Footprint: 34.65' x 34.65'
• Size: 1200.6sf each floor, 2401sf total
• Wall height: 8'
• Basement (zones 1-3): 4' above, 4' below grade, R-0.06, R-0.67
• Flooring (1st floor): 3/4" hardwood
• Skirt (zones 4-5): 280sf, R-0.68 (1" stucco), ∆-15°F vs. outside
• Attic: R-19.5 (don't laugh, I've bought homes that had less)
• Walls: R-9.3, 2012sf
• Windows: R-1, 126sf
• Doors: R-4, 80sf
• Duct losses: 10%
• ACH50 7.5 = ACHnatural (zones 1-5): 0.56, 0.43, 0.48, 0.56, 0.56
I doubt your math is wrong your assumptions are wrong
There is no surface the that has a lower R value than the air layers involved, everything starts at about 1 then you add the insulating value of the materials in question
All the materials
You have an r value of 9.3 for a wall, I assume r11 or r13 insulation with 2x4 on 16?
value for inside air layer
value for sheetrock
value for sheathing
value for siding
value for exterior air layer
when you get up to R50 these details tend not to matter so much, since they are a fixed fairly low value, but with low wall r values they add up
Yes, the R-9.3 is the whole-wall R-value for 2x4 on 16" centers.
Which of my assumptions are wrong?
Michael
I explained some of the assumptions in my post above.
There are people here, not me, who make their living studying these things, and assuming they do not know what they are talking about is a bad place to start.
A book that I still reference, while outdated an hippyish is:
'From the Ground Up' Cole and Wing
Out of print but my wife found a copy to replace my coverless one
all the math is there
Some of the wall assemblies for instance are laughable
But the math is there
This is not an insult:
You have much to learn
I don't assume that people don't know what they're talking about. Indeed, the whole point of my post was to ask for peer review. I'm always prepared to learn, and I've learned tons from GBA. At the same time, if someone merely says, "You're wrong," then that's not a convincing argument.
I do not see in your earlier reply where you pointed out any specific error clearly. I couldn't quite follow, but perhaps you were saying that I should be using R-1 for the 3/4 hardwood floor instead of R-0.68?
As for a bad place to start, calling me a spammer (since deleted) was a very bad way to start, and then claiming that my article says the opposite of what it actually says, is a bad way to continue.
If you do not see what I have said, then this is part of the problem
A 3/4 piece of plywood might have a thermal resistance of R .68,. but that is not the end of the story
The air layers add another R1 to that, as I clearly stated
Also, with an enclosed crawl space, you have no wind, which increases the r value of the exterior air layer
Additionally, are you walking on plywood?
what other materials are on the floor?
an 1/8 inch of glass has an R value of about .03[concrete is about .08 per inch for instance]
without accounting for the air layer, that would mean a heatloss at a 50 degree delta T of 166000 BTU for 100 sq ft of glass [50x100/.03]
that will never happen
Attached is an image from the aforementioned book, pp 169, from 1976, this is not new information. Note he does not calculate the total assembly with studs, but I think the point is made
I have included it not for the assumption of accuracy, but to show the complexity that you are ignoring
Basements don't seem to exist down south. It is either slab on grade or crawl.
Around me (zone 5 and 6) basements are the norm, except for cottage country or remote areas, you would have a hard time finding anything on piers/skirts.
I would rework the whole section "The problems with oversizing are cost, and possibly comfort."
Modern systems modulate. If you look at the COP vs output for a modulating system like the Mitsubishi M-Series (say, https://ashp.neep.org/#!/product/34582/7/25000/95/7500/0///0 ) you'll see the COP is higher at lower outputs. For example, when it's 82F outside the COP is 6.23 with an output of 15K and 4.93 with an output of 30K.
A lot of the advice about sizing came from single stage systems that could only be on or off. With modulating systems it's less of an issue. But you still don't want them to be grossly oversized, they have a minimum output, and when the load is less than that minimum output they have to cycle on and off and they behave like old single stage systems.
The NEEP site has a feature where you can pick a heat pump, put in your zip code and heating and cooling loads and it will tell you how many hours a year it would expect to low-load cycle. That's what you should care about.
The section in question seems to already say what I think you're suggesting. Maybe it's the title of the section itself that you think should be changed?
Heating there is not a huge penalty. but cooling oversizing is problematic
The difference in pricing between various sized mini splits is enough to make me calculate accurately
An editorial comment: why talk at all about methods you admit are poor?
To demonstrate that they're poor! GBA has multiple articles with the same theme.
Omit them, and people might try to use such methods that they heard of somewhere else (or trust their HVAC contractor who uses them). Showing why they don't work gives people a reason to use a better method.
@gusfhb: If I understand correctly, you're saying that the effective R-value of any material surrounded by air is R-1 plus the R-value of the material surrounded by air. That seems plausible but I haven't been able to confirm. Certainly I haven't seen that concept in dozens of articles or hundreds of comments here on GBA, or elsewhere, and when I went looking for it in Google i came up empty. Do you have any references besides the 50-year old book you mentioned?
Google "air film r-value."
Thank you, with that I was able to find it. (My earlier search was for "air layer", not "air film".) From ASHRAE Standard 90.1 (I-P edition, not S-I edition), section A9.4.1, air film layer R-values are:
R-0.17: exterior surfaces
R-0.46: semi-exterior surfaces
R-0.61: interior horizontal surfaces, heat flow up
R-0.92: interior horizontal surfaces, heat flow down
R-0.68: interior vertical surfaces
This is a big deal, so I'm surprised that I haven't seen it mentioned before.
Based on this new (to me) information, I've lowered the upper range for heating in my article to 20 and 23 BTU/sf for southern (e.g. Texas) and almost-southern (e.g. North Carolina).