Hey Everyone. I'm planning on installing Mitsubishi mini-splits in my ICF home. I'm considering running the refrigerant lines through my wall near the outside unit, then in the basement ceiling between the floor joists of the main floor, then up the interior wall - carving away the foam - and turning into the back of the unit. (Model# MSZ-HE18NA - or similar) My question is: Is there enough clearance with 2.75 inches of ICF foam to make the 90 degree turn into the back of the air-handling unit? Or will that be too tight? Has anyone done this before? I'd appreciate any lessons learned. Thanks!

I don't have the specific experience to answer your question, but you could run the lines up into your ceiling and use their ceiling recessed unit. Gives you a cleaner install as well. I'm also consider a Mitsubishi Mini split HVAC system but I can't get over the initial cost. With such a tight and well insulated (ICF walls, spray foam in the rafters) envelope I will most likely go with a reasonably efficient ducted heat pump (Im in Phoenix). Mini splits would be ideal and the new systems can be placed in ceilings and walls, but the cost is substantial. My 2 cents and good luck!

Posted By gracilism on 25 May 2016 11:54 AM
I don't have the specific experience to answer your question, but you could run the lines up into your ceiling and use their ceiling recessed unit. Gives you a cleaner install as well. I'm also consider a Mitsubishi Mini split HVAC system but I can't get over the initial cost. With such a tight and well insulated (ICF walls, spray foam in the rafters) envelope I will most likely go with a reasonably efficient ducted heat pump (Im in Phoenix). Mini splits would be ideal and the new systems can be placed in ceilings and walls, but the cost is substantial. My 2 cents and good luck!

Thanks for your insight. I'm surprised that mini-splits would be higher than a duct-ed heat pumps. (I'm in MD)... I always thought it was the opposite, but honestly I've never quoted a heat pump. With regard to your suggestion of putting the air-handler between the ceiling joists, that will work for one of the handlers, but for the one in the great room I have a cathedral ceiling and that's where I'm hoping I can run the lines inside, and turn into the back of the unit. Basically, I'd like to make a straight turn down, out of the back of the air handler with less than a 3" radius.

It depends entirely on what the line sizes will be. I don't remember what the line size is, but I did exactly that in my garage for a 9000 Btuh Daikin wall unit. There was no problem with bending the lines from the foam into the unit.

One thing to remember - put a sheet metal protective cover over the the lines. 16 ga works good, thin enough to not cause a problem with the wall board and thick enough to keep from driving a screw or nail through. Don't ask why I know it needs to be done!

If you haven't already done so, take a look at the Daikin units. That's what I have. I don't know if the Mitsubishi uses a variable speed compressor or not, but the Daikin does. It works super well in adjusting the heating/cooling output to match the load in the house. My unit cycles once a day normally. On in the afternoon or evening, and off in the morning. It just changes speed as needed to match the load.

dmaceld! THANKS ... Great point about covering the line set with metal / protection. I'lll take a look at the Daikin's. That said, the HVAC guy that I'm hating the least right now is a Diamond Mitsubishi Installer... (Which means he gets to officially ignore the installation specifications from Mitsubishi - and if anything goes wrong with the unit it MUST be my (select all that apply) builder's/neighbor's/dog's/engineering degree's fault). I'm meeting him at the house on Friday, I told him I'd bring the beer and the load calculations. Let's see how this goes. (At least he didn't request Miller Lite)...

(also, I'm kidding - he's a friend... are we allowed to laugh on this site at all?)

All Mitsubishis use variable speed compressors and blowers and variable refrigerant volume valving these days- Daikin has no "special sauce" on that front.

I wouldn't count on always being able to make a 2.75" radius bend in the refrigerant lines that close to the concrete, but maybe a pro could.

The HE18 doesn't have nearly as much low outdoor temp heating capacity or efficiency as the FH18. What is your 99% outside design temp, and how cold does it get during a polar vortex weather pattern? Typical design temps in MD are in the teens, but single digits aren't rare, and the HE is crapping out on capacity fast as temps fall compared to some of the higher efficiency comparables, delivering only 11.2K @ +5F (instead of 20K @ +5F.)

https://meus.mylinkdrive.com/files/MSZ-HE18NA~MUZ-HE18NA_Submittal.pdf

https://nonul.mylinkdrive.com/files/MSZ-FH18NA_MUZ-FH18NA_Submittal.pdf

Even the older Fujitsu 15RLS2 beats the HE18 it on both capacity & efficiency (both cooling and heating):

http://www.fujitsugeneral.com/PDF_06/Submittals/15RLS2%20Submittal.pdf

and the newer 15RLS3 is even more efficient:

http://www.fujitsugeneral.com/PDF_06/Submittals/15RLS3Submittal.pdf

Of course, all of this presumes you've done a fairly careful heat load calculations, and KNOW that your heat load is ~18K or less?

Posted By Dana1 on 25 May 2016 03:21 PM
All Mitsubishis use variable speed compressors and blowers and variable refrigerant volume valving these days- Daikin has no "special sauce" on that front.

I wouldn't count on always being able to make a 2.75" radius bend in the refrigerant lines that close to the concrete, but maybe a pro could.

The HE18 doesn't have nearly as much low outdoor temp heating capacity or efficiency as the FH18. What is your 99% outside design temp, and how cold does it get during a polar vortex weather pattern? Typical design temps in MD are in the teens, but single digits aren't rare, and the HE is crapping out on capacity fast as temps fall compared to some of the higher efficiency comparables, delivering only 11.2K @ +5F (instead of 20K @ +5F.)

https://meus.mylinkdrive.com/files/MSZ-HE18NA~MUZ-HE18NA_Submittal.pdf

https://nonul.mylinkdrive.com/files/MSZ-FH18NA_MUZ-FH18NA_Submittal.pdf

Even the older Fujitsu 15RLS2 beats the HE18 it on both capacity & efficiency (both cooling and heating):

http://www.fujitsugeneral.com/PDF_06/Submittals/15RLS2%20Submittal.pdf

and the newer 15RLS3 is even more efficient:

http://www.fujitsugeneral.com/PDF_06/Submittals/15RLS3Submittal.pdf

Of course, all of this presumes you've done a fairly careful heat load calculations, and KNOW that your heat load is ~18K or less?

Yep. I used the wrong model number. It's an 18k airhandling unit, but I believe I'll have an outdoor unit that can support 27k, or 36k, depending upon what the pro tells me. Primarily (I didn't get into this) the unit will be used for cooling, not heating. We have radiant floor heat in the basement, and an air-tight stove on main floor for heating. Also, this is only one unit, there is an additional 9k unit in another location in the house. If I remember right, I think I used 15 degrees for my 99% temp on the load calcs. I haven't done a Manual-J on all the rooms but I did do a "whole house" number and had it verified by BORST's model and another HVAC Pro. Everyone came up with about 29 - 31k BTU's I actually posted my calc's in another thread a while back. I used the High-Mass wall calculator. I really hope the HVAC guy insists on doing a manual J for everything. With regard to the installation, I'm concerned about that tight radius for sure, if I need to I can blast a hole in the wall, but I'd rather not. Thanks Dana1...!

Careful what you hope for- HVAC guys are notoriously all over the place on their Manual-J calculations! :-)

An aggressive cooling load calculation would be called for. You probably don't need 3 tons of cooling, and maybe not even 2.5 tons.You're probably looking at a MXZ-2C20NAHZ or MXZ-3C24NAHZ and a couple of appropriately sized heads, (or two separate mini-splits.) You would only bump up to the 3 ton multi-split if you were looking for it to cover 100% of the heat load at polar-vortex temps. (Even the 2-tonner is good for 25,000 BTU/hr @ +5F.)

http://www.ecomfort.com/manuals/MXZ-2C20NAHZ_Submittal.pdf

https://meus.mylinkdrive.com/files/MXZ-3C24NAHZ_Submittal.pdf

http://www.mitsubishipro.com/media/989222/mxz_h2i_productlaunchbrochure1stprint.pdf

Posted By Dana1 on 25 May 2016 04:08 PM
Careful what you hope for- HVAC guys are notoriously all over the place on their Manual-J calculations! :-)

An aggressive cooling load calculation would be called for. You probably don't need 3 tons of cooling, and maybe not even 2.5 tons.You're probably looking at a MXZ-2C20NAHZ or MXZ-3C24NAHZ and a couple of appropriately sized heads, (or two separate mini-splits.) You would only bump up to the 3 ton multi-split if you were looking for it to cover 100% of the heat load at polar-vortex temps. (Even the 2-tonner is good for 25,000 BTU/hr @ +5F.)

http://www.ecomfort.com/manuals/MXZ-2C20NAHZ_Submittal.pdf

https://meus.mylinkdrive.com/files/MXZ-3C24NAHZ_Submittal.pdf

http://www.mitsubishipro.com/media/989222/mxz_h2i_productlaunchbrochure1stprint.pdf

I hear ya! The extent of the calculations that some provided me were "How many square feet is your house? You need 4 tons..." Dana1, I really appreciate your help on this and other sites. Now, if we could just lighten the mood around here a bit more. :) ICF'ers are a pretty boring bunch. :blink:

One thing to take into consideration is the efficiency of the outdoor unit at partial loads. I don't know how it plays out on the cooling side, but I got stung on the heating side of the system. My outdoor unit is 36k rated. I have one large indoor unit for the house rated at 36k and one 9k unit in the garage. What I discovered is if the 9k unit is the only one running, and the outdoor temp is down to around 20° or less, the efficiency of the outdoor unit drops down to about 1. In other words, for heating if the 9k unit is the only load the system efficiency is no better than straight electric resistance heating. If the 9k unit plus the 36k unit are both running the system efficiency will be at its max at something like 300%. In my case the probability of running the garage unit at the same time as the house unit, i.e., mostly night time, was low. The probability of wanting to run the garage unit when the house unit was not, i.e., daytime hours, was high. The end result is I almost never run the garage unit and instead use a pellet stove for heating the garage.

I suggest you take a hard look at the engineering data for the Mitsubishi units in AC mode to see how the system efficiency is affected if only your 9k unit is running. Factor in a guesstimate about the percentage time it will be running by itself, and evaluate whether or not you really want to use a small unit, or whether you can have the larger unit fill the need. Also, take a look at using several smaller units. This may give you the advantage of several of them running at the same time thus keeping the overall system efficiency at higher levels.

It's often better to go with individual multi-splits than a multi-split for those types of reasons. For instance:

The minimum modulated output of most multi-splits is 6000-7000 BTU/hr (in heating or cooling), usually about the MAXIMUM output of a half-ton ductless head. A 3/4 ton FH09 can modulate down to 1700 BTU/hr, which is well below the minimum output of any MXZ-series multi-split compressor. (The FH09 can also deliver 12,000 BTU/hr of cooling when it's 95F outside, even though it's efficiency is tested at the "rated" 9000 BTU/hr). If the cooling load at the 1% outside design temp is under 12K, it's often a better choice than the 1-ton, which only modulates down to 2500 BTU/hr in cooling mode.

Fujitsu equipment has the same issue- all of their single-head units modulate down to 3100 BTU/hr, but the multi-split compressors at min-modulation are ~2x that much.

If the average load is below the minimum modulated output of the compressor and cycling on/off there's no way it can meet it's efficiency numbers. But with a pair of mini-splits you can turn one of them completely off when loads are low, extending the modulation range downward. If you oversize the hell out of the compressor by 2x just to get more zones out of a multi-split it's even worse- the compressor needs a minimum amount load to oparate at it's potential super-efficiency at part load.

Let me guess- if they said you need 4 tons, it means you have between 2000-3000' of conditioned space? Tbe BTU/foot hacks typically use a ton per 500' , or sometimes a ton per 750'. Realistic Manual-J loads on code min new housing usually runs about a ton per 1000', but can easily be a ton per 1500'. There are exceptions to the rule (say, some huge "sunset view" floor to ceiling west facing windows), but that's why you run the numbers rather than some idiot rule of thumb.

Oversizing modulating equipment to the point that it can't modulate should be crime...


...but that would put half the HVAC installers in the country in jail.

To avoid oversizing, hire a qualified engineer or a RESNET rater (and NOT an HVAC contractor) to do an aggressive room-by-room Manual-J on the house, using extremely aggressive assumptions on air-tightness and assume that it is using heat recovery ventilation at minimum ventilation rates, and 68F & 80F as the indoor heating & cooling design temps. Only when you know the loads can you properly specify the equipment. Even if it costs you a grand for the third-party Manual-J you'll save more than that on equipment, as well as on the operating cost down the line.

Dana, what are you thoughts on the new VRF systems from Trane/American Standard and Lennox (and Mitsubishi's new ducted air handler set up)? Could those variable compressors solve a potential oversized issue? Of course you still wouldn't want to over buy, but could they at least act like smaller more efficient systems? I've heard that the Trane XV18/20i can modulate from 0.1 to 5 tons.

Posted By gracilism on 25 May 2016 11:54 AM
I don't have the specific experience to answer your question, but you could run the lines up into your ceiling and use their ceiling recessed unit. Gives you a cleaner install as well. I'm also consider a Mitsubishi Mini split HVAC system but I can't get over the initial cost. With such a tight and well insulated (ICF walls, spray foam in the rafters) envelope I will most likely go with a reasonably efficient ducted heat pump (Im in Phoenix). Mini splits would be ideal and the new systems can be placed in ceilings and walls, but the cost is substantial. My 2 cents and good luck!

In the PHX area one can get a Mitsubishi ductless mini-split (3/4 - 1 ton) 28 SEER unit for about $3,000 installed, warrantied for 10-years. That's VERY inexpensive compared to ducted units which can easily go into the $25k range once you add the duct install labor, returns, air handler, etc.

Once you introduce duct work, efficiency drops, that's the reality of HVAC physics. Most AC companies screw up duct work, that's the other reality. Ducts are undersized, crazy bends and turns, leaky ducts, oversized AC units, etc.

Best thing to do is install a 3" round PVC protrusion into the ICF wall. This is where the wall mount mini would go and the refrigerant, electrical and drain lines will easily fit through that 3" opening.

In Phx you are in a cooling dominated climate. A heat pump is all you will need. Radiant heating is complete and total overkill. Just stick with a few strategically placed ductless mini splits with heat pumps and you are good to go. I would stay away from Trane or American Standard and any ducted unit for that matter.

Posted By gracilism on 26 May 2016 05:36 PM
Dana, what are you thoughts on the new VRF systems from Trane/American Standard and Lennox (and Mitsubishi's new ducted air handler set up)? Could those variable compressors solve a potential oversized issue? Of course you still wouldn't want to over buy, but could they at least act like smaller more efficient systems? I've heard that the Trane XV18/20i can modulate from 0.1 to 5 tons.

Show me the data on that modulation range! (Could't find it quickly on the Trane website.) Let's just say I'm skeptical that a 5 ton compressor can modulate down to 1200 BTU/hr efficiently. Is there an engineering manual online you can point me to?



Most split systems with larger air handlers have less than a 3:1 turn down ratio. The Carrier GreenSpeed at low speed is still delivering 40% of the max, which is a 2.5:1 ratio. I'm sure that can be improved upon, but 60,000 BTU/hr down to 1200 BTU/hr would be 50:1, which sounds like something the tooth fairy might spin while riding by on a unicorn. :-)

A 3-ton Mitsubishi with the 3-ton PVZ seriseair handler can deliver 29,000 BTU/hr heating @ +5F, but it's only a 2.2:1 turn down- min output @ 47F is 18K, max is 40K.:

http://www.ecomfort.com/manuals/mitsubishi-fdac2a28c70cd8b8d70f99f3c7c7d934.pdf

I'm not sure what the 3 ton MVC air handler's minimum modulated output is when married to the 4 ton compressor, but it'll deliver 40K @ +5F. Don't know what it's minimum output is but I suspect it's on the order of a ton, possibly more.

Im sure you're correct. That info came from a local subcontractor and was "off the top" of his head.

... or maybe he heard it directly from the tooth fairy! :-)

Posted By Dana1 on 26 May 2016 05:03 PM
It's often better to go with individual multi-splits than a multi-split for those types of reasons. For instance:

The minimum modulated output of most multi-splits is 6000-7000 BTU/hr (in heating or cooling), usually about the MAXIMUM output of a half-ton ductless head. A 3/4 ton FH09 can modulate down to 1700 BTU/hr, which is well below the minimum output of any MXZ-series multi-split compressor. (The FH09 can also deliver 12,000 BTU/hr of cooling when it's 95F outside, even though it's efficiency is tested at the "rated" 9000 BTU/hr). If the cooling load at the 1% outside design temp is under 12K, it's often a better choice than the 1-ton, which only modulates down to 2500 BTU/hr in cooling mode.

Fujitsu equipment has the same issue- all of their single-head units modulate down to 3100 BTU/hr, but the multi-split compressors at min-modulation are ~2x that much.

If the average load is below the minimum modulated output of the compressor and cycling on/off there's no way it can meet it's efficiency numbers. But with a pair of mini-splits you can turn one of them completely off when loads are low, extending the modulation range downward. If you oversize the hell out of the compressor by 2x just to get more zones out of a multi-split it's even worse- the compressor needs a minimum amount load to oparate at it's potential super-efficiency at part load.

Let me guess- if they said you need 4 tons, it means you have between 2000-3000' of conditioned space? Tbe BTU/foot hacks typically use a ton per 500' , or sometimes a ton per 750'. Realistic Manual-J loads on code min new housing usually runs about a ton per 1000', but can easily be a ton per 1500'. There are exceptions to the rule (say, some huge "sunset view" floor to ceiling west facing windows), but that's why you run the numbers rather than some idiot rule of thumb.

Oversizing modulating equipment to the point that it can't modulate should be crime...


...but that would put half the HVAC installers in the country in jail.

To avoid oversizing, hire a qualified engineer or a RESNET rater (and NOT an HVAC contractor) to do an aggressive room-by-room Manual-J on the house, using extremely aggressive assumptions on air-tightness and assume that it is using heat recovery ventilation at minimum ventilation rates, and 68F & 80F as the indoor heating & cooling design temps. Only when you know the loads can you properly specify the equipment. Even if it costs you a grand for the third-party Manual-J you'll save more than that on equipment, as well as on the operating cost down the line.

I've left room (added penetrations) for up to three outdoor units... We'll see what the Manual-J says. I ran through the concept - including ventilation and circulation with the HVAC guy on Friday. I asked him about modulating the the larger multi-split outdoor unit down so that the 9k airhandler -if it were running by itself- would operate efficiently. We're going to check the numbers and evaluate. Thanks all for the great discussion.