Minisplit sizing dilemma — to replace oil furnace
I love the GBA website. Thanks for so much good information.
We’re in the planning phase for replacing the 31 year-old oil furnace in our 1200 sq. ft. house in Portland, Oregon (climate zone 4C) with a mini split heat pump. The heat pumps we’re considering is the Daikin FTQ18PVJU or the FTQ24PVJU.
The house was built in 1960 and is 3-bedroom, 1-bathroom, 2-story, slab-on-grade construction. Windows have been replaced with double-pane, low-E windows and modern solid-core wood doors with tight weatherstripping but those are the only energy improvements the house has installed.
Current furnace: Lennox model O12Q3-105-3, installed in 1987. Furnace is fired at 0.65 GPH. Annual fuel consumption averages 230 gallons per year over 15 years.
I did a Manual J load calculation at loadcalc.net which results in a heating load of 23,412 BTU per hour at 68 deg F indoor and 24 deg F outdoor conditions.
As a sanity check, I compared this against the last severe winter weather we had here in January 2017 that lasted 10 days and saw average daily temperatures of 27.8 deg F and temperatures as low as 11 deg F according to the National Weather Service. The ACCA Manual J design temperature for Portland is 24 or 27 deg F (there are two values listed for Portland), close to the average over the 10 days of this cold snap.
Our furnace ran 3.94 hours per day average over those 10 days. At 0.65 GPH that means we consumed 2.56 gallons per day average. Assuming the furnace is 80% efficient that means we needed 284,000 BTU per day. That’s 11,833 BTU per hour.
During this cold snap we kept the house at 68 deg F from 6:30 AM to 10:00 PM (15.5 hours per day) and 60 deg F from 10:00 PM to 6:30 AM (8.5 hours per day) for a 65 deg F average indoor temperature.
So I went back to my Manual J calculations and changed the interior temperature to 65 deg F and the outdoor temperature to 27.8 deg F. Under those conditions it predicts a heating load of 19,794 BTU per hour. That’s 67% more that what the furnace run-time suggests was our actual heat load.
I understand that oversizing heat pumps is bad news, so I’m trying to get this correct. There’s obviously a big discrepancy between these two calculations and that’s confusing me.
Thanks in advance for any help you can give me in selecting the correct system capacity.
GBA Detail Library
A collection of one thousand construction details organized by climate and house part
Replies
There is too much "noise" in short duration measurements of fuel use or duty cycle to come up with reasonable accuracy. Oil fill-ups that take place during the winter with total heating degree-days between fill-ups works if most of that fuel use was during the winter.
For the 10 day cold snap period, the implied heat load from fuel use works this way:
At a presumed 65F heating/cooling balance point is (65F - 28F=) 37F heating-degrees.
The average 11,833 BTU/hr @ the average ~ 28F is then 11,833 / 37= 320 BTU per heating degree.
The 99% outside design temp for Portland is 24F (27F at the airport), which is (65F-24F= ) 41F heating degrees, an implied load of 320 BTU/degree-hour x 41 F heating degrees = 13,120 BTU/hr @ 24F.
That's your approximate heat load as MEASURED, not 23K, as calculated from the Manual-J.
The greatest amount of oversizing you'd ever need is 1.4x, , which would be 18,368 BTU/hr, which is within the 20,000 BTU/hr capacity @ +17F specified in the submittal for the Daikin FTQ18PVJU:
http://www.daikinac.com/content/assets/DOC/SubmittalDataSheets/SDS%20FTQ18PBVJU_RZQ18PVJU9.pdf
If there is a 1.25 tonner in that series it might be more appropriate for your load than the 1.5 tonner.
[edited to add]
Looks like the 1.5 ton is the smallest in the series, and more than enough capacity for your loads:
http://www.daikinac.com/content/assets/DOC/Product%20Brochures/CB-FTQDUCTED.pdf
Even though a 10 day run is very short with a lot of potential for measurement error, it won't be off by more than 25%. Even if rather than 13K @ 24F it's comes in at (1.25 x 13K=) ~16K, the FTQ18 still covers it with a ~1.25x oversize factor, with is fine.
A similar mini-split that might be more appropriately sized for a ~13K load is Fujitsu's 1 ton 12RLFCD, which is good for 16,000 BTU/hr @ +17F:
http://www.fujitsugeneral.com/us/resources/pdf/support/downloads/submittal-sheets/12RLFCD.pdf
But if the idea is to marry it to the existing duct system the bigger air handler that comes with the FTQ is probably going to work out better than the RLFCD.
Looking at the specs of these 2 model #'s, it seems likely these are the same pieces of equipment, with slightly different controls.
Other than the windows, have you done any specific air sealing? Have you had a Blower Door Test? A tight house heats much easier and much less expensively than a drafty house.
Dana, thanks for the great answer. I like your BTU per heating degree day calculation method. I also found your response to this article: https://www.greenbuildingadvisor.com/blogs/dept/musings/how-perform-heat-loss-calculation-part-1
where you demonstrate a BTU per heating-degree-hour calculation, which I really like.
I agree with you that there's a lot of noise in short period measurements and that's a weakness of my 10 day measurements & calculations. Unfortunately, we don't have oil fill-ups during the winter heating season. Our tank is 675 gallons and we only fill it up once every 2-3 years, usually in the summer when prices are lowest.
What I do have access to is heating oil purchase records going back to 2002. So I used those records and the heating degree days between purchases to follow your calculation methodology. Between August 2002 and March 2016 we averaged 264 BTU per heating-degree-hour (base 65 deg F). That corresponds to 11,096 BTU/hr for 23 deg F outside and 65 deg F inside.
Also, our last tank fill-up was March 2016 and in July 2017 I measured the oil level in the tank using the calibrated stick provided by our heating oil supplier. Using that oil consumption, along with the heating degree days between those dates, I calculate 301 BTU per heating-degree-hour (base 65), or 12,640 BTU/hr at 23 deg F outside & 65 deg F inside.
So I now have three calculations covering different time spans and using different methods:
10 days in January, 2017 using furnace run-time: 12,583 BTU/hr
14 years oil purchases, 2002-2016: 11,096 BTU/hr
15 months oil consumption (tank level measurement), March 2016 - July 2017: 12,640 BTU/hr
The average of all three methods is 12,106 BTU/hr. Each of the individual calculations are within 8% of the average. That gives me more confidence that these calculations from measurements are giving me reasonably accurate sizing estimates for the new mini split. After I correct each of these for a 68 deg F interior temperature the average becomes 13,220 BTU/hr, which is still within the capabilities of the 1.5 ton system.
I will check out the Fujistu 12RLFCD. We are planning to use the existing duct work, so you're right, the Daikin's larger air handler may work out better.
Thank you again for your help.
John, thanks for your reply. If the only difference is the controls, does that mean I might be able to "upgrade" the 1.5 ton system to a 2 ton system if I find I need more heat?
Bob, thanks for your reply. No, we haven't had a blower door test or done much in the way of air sealing. Aside from the windows & doors the construction of the house is pretty typical of a house built in 1960. I know there's R-11 insulation in most of the wall cavities (though not all, as I found out when a water pipe froze one winter in an uninsulated cavity). The ceilings are cathedral and I don't know what the insulation is like. I haven't found a way to inspect the ceiling insulation aside from cutting holes somewhere.
Among similar bigger air handlers, Mitsubishi's MVZ-A12 puts out 13,500 BTU/hr, which may be cutting it a bit close. Their MVZ-A18 is very comparable to the 1.5 ton Daikin. I believe they're only compatiable with their mulit-zone compressors though:
https://resource.gemaire.com/is/content/Watscocom/Gemaire/mitsubishi_mvz-a12aa4_article_1436343373152_en_ss.pdf?fmt=pdf
Dana, thanks, I will investigate the Mitsubishi MVZ-A12 also.
What is the risk from over-sizing the mini split heat pump by more than 40%?
The modulated output range of most of the bigger-deal air handler units it typically less than a 3:1 turn-down from maximum/minimum. So if your design load is 13,000 BTU/hr and the maximum output is 20,000 BTU/hr at minimum-modulation of 20,000/3= ~6,700 BTU/hr is more than half your design load, and may even be above your AVERAGE load in mid-winter.
To get the most efficiency and comfort out of a modulating heat pump it has to be modulating most of the time rather than cycling on/off. Every time the compressor has to spin up from zero there is an associated amount of power used that is never recovered. Ideally the cycles in mid-winter would be extremely long, nearly continuous, and that doesn't happen if the minimum output is more than half the design load.
Dana, thanks again but I'm a bit confused by the Daikin engineering data and I'm concerned about having too much capacity with the 1.5 ton air handler.
I've been reading Daikin's engineering manual: http://www.daikinac.com/content/assets/DOC/EngineeringManuals/2017/Engineering%20Manual-SkyAir-web.pdf
On page 27 it shows the heating capacity as 20,000 BTU/hr @ 47 deg F outdoor dry bulb and 13,000 BU/hr @ 17 deg F dry bulb. It also shows the capacity control as being 35% to 100%.
However, when I look at page 98 it shows the capacity as 20,740 BTU/hr @ 14 deg F to 50 deg F outdoor wet bulb and 68 deg F indoor dry bulb.
Now I'm confused as to what the actual capacity at 17 deg F outdoor really is. If it's 13,000 BTU/hr then that's just about right for my house. But if it's 20,740 BTU/hr then it's 59% over-sized. Now I'm concerned that I may have comfort and efficiency problems with that unit, based upon your warning about not exceeding 40% over-size.
Assuming 20,000 BTU/hr capacity and a 35% control point then the minimum output is 7,000 BTU/hr. If the house requires 13,000 BTU/hr @ 23 deg F outdoor and 0 BTU/hr @ 68 deg F outdoor then linear interpolation says the 7,000 BTU/hr minimum capacity balance point would be ~44 deg F. Does that seem acceptable or is it too low?
Do you have insight on this? Am I misinterpreting what I'm reading the engineering manual?
Thanks again for your help.
At any set of wet & dry bulb temperatures it has about a 3:1 turn down ratio. (35% is about 1/3 of 100%), but the capacity changes with both dry bulb and wet bulb temperatures- it's lower at +17F than it is at +47F. The wet & dry bulb temperatures at which the unit is tested for HSPF efficiency are prescribed conditions of the test. To be able to state a nominal heating capacity it has to be able to deliver at least that much at +17F, and be able to modulate down to that level at +47F when testing it's HSPF efficiency.
It's minimum output at +47F is about 7000 BTU/hr, and your house's load increases at 320BTU/hr per degree F below 65F/ So your load crosses 7000 BTU/hr at only 7000/320= 22F less than 65F, or about 43F, which is roughly your mean temperature in January. see:
https://weatherspark.com/m/757/1/Average-Weather-in-January-in-Portland-Oregon-United-States#Sections-Temperature
and
https://weatherspark.com/y/757/Average-Weather-in-Portland-Oregon-United-States-Year-Round
That means it'll still be doing a lot of cycling even during the coldest month. Ideally you'd have something that can modulate down to about half that much (or less), so that it will modulate-mostly even during the shoulder seasons. The Fujitsu 12RLFCD and 18RLFCD mini-duct units both modulate down to 3000 BTU/hr (which means it would be modulating with load even in the low 50s F outdoors) but the duct design needs to really work with the smaller air handler.
I suspect the Mitsubishi MVZ-A12 might be the most reasonable fit among the bigger air handler mini-splits.
I've been doing similar calculations trying to decide whether a 36,000 BTU Mitsubishi PUZ Hyper Heat heat pump would be sufficient for our 2,250 square foot house, as one contractor recommends, or whether we would be better off with the 42,000 BTU one (as the contractor who actually did a Manual J calculation recommends). I used approximately the same time period (December-January heating bill from January 2017) as it was the highest bill in recent memory. If I assume our furnace is 80% efficient (as it originally was supposed to be, when installed a couple of decades ago), I do end up with a value only a bit under 42,000 BTU/hour (after applying a 1.4 fudge factor as Dana's method suggests). If I assume it's closer to 60% efficient, I get about 31,000 BTU/hour. But one wrinkle that I didn't notice at first is that the stated capacities of the Mitsubishi models are their cooling capacities, while their heating capacities are higher: the "42,000 BTU" model is actually rated at 54,000 BTU for heating (the Manual J suggested a value between those numbers). However, the 36,000 BTU model appears to be only 40,000 for heating -- perhaps it doesn't matter as it would mean about a 1.36 fudge factor compared to 1.4, and that only if we assume the current furnace is operating at anything like its original efficiency, which seems unlikely.
It's of course possible that the Manual J is off due to things like the preferred temperature being set too high (it was at 70 when in fact we keep it at 66, 55 at night -- yes, I know we would probably not want a setback that large with a heat pump), the heat loss from the windows being overestimated, etc. (We do intend to replace some windows, but mostly for aesthetic reasons -- only two of the original single pane leaky windows are left.) But I don't know how big an effect those things would have.
IRENE3: What is your outside design temperature?
Dana -- it's 26 degrees.