I always see mini-splits mounted on an outside wall- Is there any reason you can't mount the inside unit on an interior wall? The drain and refrigerant lines could run down inside the wall, then outside through the basement. Although the runs would be a bit longer, the lines would be inside semi-conditioned space.

Also- Is it better to run a multi-head system, or individual units? The individual ones have MUCH higher SEER ratings.

You can employ one anywhere you can communicate condensate and refer lines to destination. Mini condensate pumps even allow you to move condensate up hill if necessary.
Amongst the efficiencies of mini splits is the simplicity. The more elaborate you get the "less efficient" you will be. Adding heads, or ducts or fans (to communicate to different spaces) all have friction or line loss or added amp draw.
This accounts for some of the "inefficiency" of ducted systems.
There has been a lively debate about this, and again I'll throw another request for an ASHP forum to the mods, adding a bunch of components to a mini split means some of it's proponents are swiftly moving towards inventing a ducted hvac system with all the attached inefficiencies. Much like the folks who suggest oversizing to keep equipment in low stage is more efficient (inventing single stage equipment all over again) thus tanking operating cost by running a larger unit all the time.
Every product has an application for which it is most effective. Few products work well in every application.

Before you start making decisions based solely on the SEER ratings, you should take a close look again at what the SEER numbers really mean. In general, as the output of a unit goes up, the SEER rating will go down. Larger units are rated at full output, but that isn't necessarily how they will actually run over the entire season. You might want to figure out what energy savings your situation would generate with a 25 SEER unit vs a 20 SEER unit, for example, and then use that information to help decide what measures are actually worth it. I find it hard to believe that having four different locations for outside wiring, compressor noise, exhaust, appearance, etc would be preferable to one. I'm sure individual units come with some sort of a price premium as opposed to a system. "Green" considerations would entail looking at additional construction materials and energy required, packaging, freight and installation issues, not to mention refrigerant charge.

All my mini-split heads are mounted on interior walls, including one that is 5 feet below grade. The linesets are adequately insulated, so there is no difference as to running them through conditioned space or not. Drain lines are small enough to run right through stud walls (to the outside) just like any other plumbing or wiring.

Mounting the head on interior walls is fine, but it can complicate the refrigerant line runs.

Being fully modulating (not simple multi-stage) systems with fairly high turn-down ratios, the as-used SEER will vary by how big the compressor is relative to the load- a bigger compressor & coil relative running at part load delivers much higher efficiency than at rated load. The loads of the different zones of a multi-split won't necessarily track each other (the west side load will peak in the PM, the east side in the AM, etc.) and evenly-split output between heads used for rating the system for SEER is likely to be closer to a worst-case scenario, and the phase difference between zones means that the zone that's peaking in load is getting the benefit of the outdoor coil & compressor's oversizing at least most of the time. Real world SEERs are more likely to exceed the manufacturer's spec for multi-splits.

Oversizing the outdoor coil for the individual or combined zone-loads is in no way comparable to Joe's "inventing a single stage equipment all over" comment since these aren't simple multi-stage systems. Rather, these systems have continuously variable refrigerant volume control, as well as continuously variable compressor speeds & blower speeds, that simply take advantage of the fact that the outdoor coil is oversized for the part-load conditions, and the component speeds & refrigerant volume can (and is) adjusted to run maximally-efficiently for the actual load as it varies in time. The output range between minimum & max can even exceed 5x, but the highest efficiencies happen at the lower output end, until the load is so low that it's cycling on and off (which takes a toll on efficiency) rather than modulating. (As would be the case with Joe's oversized multi-stage unit examples that are cycling on/off, even if they're only running on at the low stage.)

The part load efficiency depends on both the interior head's coil size relative to the load of that zone and critically, the control algorithm embedded in the firmware that adjusts the blower & compressor speeds and the refrigerant volume, not just the compressor & outdoor unit's coil size relative to the load. The algorithms for controlling multiple heads with a single compressor aren't that different from a single-head, but with the compressor running at a speed that supports the multi-zone load it's going to be (very) slightly less optimal than with a single-head, single zone. They're getting better every day though- testing in the Pacific Northwest under the NEEA's ductless pilot program shows that newer model years of exactly the same systems have higher measured in-situ efficiency than presumably identical systems only 2-3 years older. The "special sauce" that differentiates them is in the control firmware, since all of the mechanical & electrical components are nominally the same, fully interchangeable.

The efficiency hit of multi-splits are also in no way comparable to the inefficiencies of ducted systems, as Joe would have it. The refrigerant lines aren't necessarily going to grow in length to the point where the head presented has a measurable effect on compressor efficiency- most of the backpressure is from the valving in the interior head. This is quite a different problem than the friction aspects of longer ducts. And since the interior head is only recirculating the air within a room, it doesn't create pressure differences between rooms to drive outdoor air infiltration the way ducted systems do. (Even the best-balanced best-sealed duct systems create some infiltration drives, which reduces the net efficiency.)

It's very rare that a house would need a head in every room to maintain comfort, but optimizing the location of the head within the zone it's trying to cool or heat takes some consideration. Start with a room-by-room Manual-J type calculation of the loads as a start, and consider how much (non-powered, only convection) air-communication there is between adjacent areas. Placing the head near the biggest load helps. Many homes will do just fine with one head per floor, placed in the room or area with the highest calculated load, others may have too many high-gain rooms that are doored off from one another, and may need more heads. As the number of heads clime the upfront cost can be pretty substantial, but a 4 head 3 ton multi-split is (usually) still cheaper than four 3/4 ton mini-splits.

Our house should be pretty easy to condition; there are essentially only two rooms! The entire back of the house is basically one big room, divided into thirds by a vestigial wall, only enough to give the appearance of a separate room. Maybe 16" wide on each side, and a lowered header from the ceiling. Cosmetic trim will accentuate the effect. The kitchen will be on one end. the LR on the other, with a dining area in between. I figure one head in the kitchen and one in the LR would suffice. The front of the house consists of a foyer, powder room, stairwell, and a closed-off master bedroom, which would also get its own head.

The entire main floor is only 1250 sq , ICF, and the attic will be R-48 or better. The climate is upstate SC, with about 3 months of really hot weather, and a mild winter. We keep our A/C at 80 in the summer, and the heat is set around 55-60, so we are light users. We use ceiling fans a lot, too. This is on top of an ICF basement. Having 3 heads on one unit would work, although the refrigerant runs would get longish- up to 35' or so. Maybe it would be better to have the bedroom on its own?

Do the room by room load calculations- it makes a difference.

In an ~R22-ish ICF building with an ~R50 attic, solar gains from windows and shading factors have the biggest effect on peak loading. It's not inconceivable that a single head could handle the entire first floor if placed in the room with the largest window gains.

Orientation matters- what cardinal direction is the "back" of the house facing?

Most ductless manufacturers are fine with refrigerant line runs of up to 50' (and even that can be pushed a bit) but installation can be a PITA if it has to make 18 bends along the way.

If the basement is mostly below grade and doesn't have much window gain to be concerned with you may not need to cool the basement as long as the first floor is maintained at temp. But if you're heating with the ductless too it will need it.

In a house that is occupied more than 12 hours/day it's usually more efficient (= uses less power) to "set and forget" the temperature on the ductless and let the system modulation automatically track the load at highest possible efficiency. A set-back strategy generally uses more power because it's efficiency is dramatically lower running at max speed during the recovery ramp than it is when idling along tracking the load.

Having a separate zone dedicated to the bedroom only makes sense if you need to run the bedroom several degrees cooler than the rest of the conditioned space (which is often the case.) Installed costs are on the order of ~$1.5K/head for the interior units, plus the base ~$2.5K/ ton for a better-class high efficiency 2-3 ton multi-splits. You can see how it gets out of hand when going for large numbers of heads! But there are decent 8-head 4-ton units out there, if you REALLY want them. (Those Medusa systems are more commonly installed in commercial buildings than residential.)

Some mini-splits won't go as low as 55F in heating mode, but most can handle 60F. One aspect of these systems is very stable room temperatures which can make it somewhat more comfortable at lower heating temps. But most people would opt for the 65-70F. Since the heating efficiency would be so high in a SC type climate the difference in the monthly bill of running it 65F or 68F rather than 60F would be pretty minimal for a 1250' house with your R-values even during January.

Without doing the actual load calculations, even using 20F as an outside design temp (23-28F is more likely) your peak heating load is probably well-under 20,000BTU/hr so running a 2 ton would likely give you huge margin on both cooling & heating. There are very few sub-2 ton multi-splits that accomodate 3 heads though. There is a wider selection of sub-2-ton dual-head units out there. Sanyo makes a pretty good 19,000 BTU/hr 3-fer (model # CMH1972) and there may be others- it's a moving target. But most 3-head units are 2 tons & up (eg. Daikin 3MXS24JVJU, Mitsubishi MXZ3B24NA or MXZ3B30NA.) Unless you have a LOT of unshaded west-facing window area your peak cooling loads are likely to be under 20,000BTU/hr too (and could be as low as half that), and going with nominal ratings as much 2x or more your actual peak loads may run less efficiently due to cycling.

Is doing a "hybrid" design (utilizing a central HVAC setup but also some rooms having mini-splits) seen as a waste of energy & money or does it depend on the home design?

The back of the house faces NNE, so there isn't much solar gain, even in the summer. The front of the house (SSW) has a deep porch across 2/3's of it, and the WSW side has an attached garage. Further, we are surrounded by tall hardwoods that block the morning and afternoon sun.

The basement wouldn't need cooling, but maybe a de-humidifier, and only modest heat, if in fact we use the basement. A friend who lives nearby said his basement runs about 55 in the winter, but is largely uninsulated. Either way, the basement ceiling will be insulated and sheet-rocked for a one hour fire rating.

What we're after is a house that is cheap to operate, but is still comfortable. It sounds like 3 heads would be ideal. I'm afraid that only one head wouldn't provide even enough conditioning in a 14' X 42' room, but then again, I've never lived in a tight, well-insulated home, either. In fact, none of the houses I've lived in were insulted, even in CT! No wonder I'm not fussy about temperatures. South Carolina has long stretches in the spring and fall where no conditioning is needed. I'm living in my barn while I build, and only the coldest days of winter and hottest days of summer are uncomfortable. I would simply have the system off during the long stretches of nice weather.

I'm afraid that only one head wouldn't provide even enough conditioning in a 14' X 42' room,

Actually doing the load calculations and comparing them to unit output might address that fear.

Every home I've ever seen has a natural circulation pattern and you need to take that into account when you place the head(s). If the house is so tight as to not have circulation, then you get to create one with your chosen ventilation plan.

Posted By jdebree on 09 Aug 2012 07:05 AM
The back of the house faces NNE, so there isn't much solar gain, even in the summer. The front of the house (SSW) has a deep porch across 2/3's of it, and the WSW side has an attached garage. Further, we are surrounded by tall hardwoods that block the morning and afternoon sun.

The basement wouldn't need cooling, but maybe a de-humidifier, and only modest heat, if in fact we use the basement. A friend who lives nearby said his basement runs about 55 in the winter, but is largely uninsulated. Either way, the basement ceiling will be insulated and sheet-rocked for a one hour fire rating.

What we're after is a house that is cheap to operate, but is still comfortable. It sounds like 3 heads would be ideal. I'm afraid that only one head wouldn't provide even enough conditioning in a 14' X 42' room, but then again, I've never lived in a tight, well-insulated home, either. In fact, none of the houses I've lived in were insulted, even in CT! No wonder I'm not fussy about temperatures. South Carolina has long stretches in the spring and fall where no conditioning is needed. I'm living in my barn while I build, and only the coldest days of winter and hottest days of summer are uncomfortable. I would simply have the system off during the long stretches of nice weather.

Based on the description your cooling loads are going to be dramatically lower than you anticipate. 

A typical SC house built between 1950 & 1980 had low-density R19 batts in the attic and R11 or R13 low denstity batts in 2x4 construction (sometimes R7-8 half-thickness "econo-batts"), and maybe a brick veneer for cladding (if you're lucky), and single pane windows with no E-coatings.  Even so it was an upgrade from houses with no insulation, and at best rosin-paper air barriers in the walls/floors/ceilings, and it was still possible to cool them using a 1-ton/500' of living space rule of thumb.

Current code-min in SC (since 2009) requires U0.65 (max) windows, which is a minimal double-pane with no E-coat, R30 (min) attic insulation, R13 mid-density batts for wood frame (R5 continuous insulation if mass-wall).  With select upgrades to minimize solar gains (including minimal "cool roof"  shingle roofing, not necessarily "white") that sort of house can typically be cooled using 1-ton/1000' rule of thumb.

As of January 2013  the IECC 2009 will be in effect, which calls out U 0.50 windows (minimal double pane, one low-E surface), but a maximum solar gain rating of  0.30 SHGC, the same R-values but requires duct leakage testing for ducted HVAC when ducts run outside of conditioned space, and whole house air leakage < 7 air exchanges per hour @ 50 pascals pressure. The true cooling loads for most code-min homes will be somewhat less than 1-ton/1000' of (above-grade) conditioned space.

In your house, with an absolute minimalist R16 ICF (no longer available from many vendors- R20-R22 is the minimum for many) and your R50 roof you're roof gains will be half that of a code-min house, and with the mass-wall the peak wall gains will be less than 1/4 that of a 2013 code-min house, and it will be almost ALL window-gain.  Even with code-min windows, given your orientation & shading factors, the odds are your peak cooling loads will be less than 10BTU/ft, including internal gains.  There are many interior-heads rated for 15,000BTU or more, but even a 9000BTU head might actually cover it (which would be 7.2BTU/foot for a 1250' house.)

Most mini-splits have a "dehumdify" mode, but SFAIK all but the Daikin Quaternity series achieve that simply by running the interior head coil at very low temp with minimal air flow to max out the condensation, and do NOT run it under dehumidistat control.  By contrast the Daikin Quaternity series has independent relative humidity and temperature setpoints, and is capable of dehumidifying even in heating mode (whereas all others will be cooling while in dehumidify mode.)  These are high-end units and you can expect a higher installed price, but the premium may be worth it if precise humidity control is important to you. 

SFAIK the Quaternity only comes in single-head versions, not multi-splits, but Daikin does have a decent range of multi-split offerings (up to 8-heads). Their smallest 3-fer is a 2-ton, but can be used with heads rated as low as 7KBTU/hr (which is probably the right sized head for your basement zone) or as high as 18KBTU/hr, with a range in-between, all EnergyStar rated. They also have a 1.5 ton 2-fer that might work if your load analysis comes in under 9KBTU/hr for the first floor (the largest head it accommodates is a 9K.) See: http://www.daikinac.com/residential...ts&page=55

But there are lots of good units out there, and it's a moving target.  The key comfort & success is to do a real load calc, then see what's out there that works for you at a reasonable upfront cost, but  I doubt you'd have to spend as much as $8K total for a super-efficient super-comfortable system that both heats & cools with some margin.

What happens when you close the door to a room that does not have an air handler unit inside it? There is no air circulation there. I am confused how mini splits even function well in a residential home where there are multiple bedrooms, closets, kitchen, study etc... and only a few air handlers to supply the whole house. I would think some areas would be devoid of cooling.

In general, closed doors in very cold weather or with solar gain don't work so well. Exactly how well depends on the variables.

OP here, and I'm happy to report that the minis are working very well, at least in the summer. Haven't been through a winter yet. I ran a Manual J program, and cooling load was just under 9K BTU, and heating was about 12K. We went for a 12K main unit, and a 9K in the master suite. The main unit does just fine for the whole house (interior doors open), and we run just the 9K at night.

We've talked at length on another thread about temperature transfer in and out of closed rooms, but I can't remember which thread. It depends upon the heating/cooling load for the room, and how tight the doors are. Since interior walls are generally not insulated, some temperature transfer will go through the walls. There are various ways, such as active or passive vents to circulate the air. Our situation is not typical, as we only have one bedroom that we use regularly, our house is a very open plan, and we always leave the interior doors open. If we have company and want more privacy, we can close our BR door, as the BR has its own head.

Do you have air flow into closets and such? I have a quote for mini splits, with ducted air into closets and bathrooms. The price tag is a bit overwhelming, and I'm wondering if it is too much. I live south of Atlanta, so cooling is pretty important. I'm wondering if I need all the interior units I've been quoted. Basically, the house is a ranch on basement, with a ridiculous 3200 sq feet of space. The quote calls for two 2.5 ton units, with some 7 different interior air handlers. It just feels like too much.

with ducted air into closets and bathrooms.

If your exterior shell is code or better, I don't see much need for putting ducted air into closets or bathrooms. What do you have for a ventilation system? Ideally, you can put ventilation intakes in closets and bathrooms. Removing smelly or humid air there allows conditioned air from a neighboring room to enter.

5 tons for 3200SF suggests a code minimum envelope.

Posted By joe.ami on 22 Sep 2014 08:22 AM
5 tons for 3200SF suggests a code minimum envelope.

The 5 tons is probably what is necessary to support that many heads, not the actual loads.   Unless this house has a square mile of west-facing windows or it's located in Antarctica 3 tons of ductless would probably be more appropriate, but it's hard to find units that small that would support that many heads.

Seven ductless heads for a 3200 square foot house feels off, by more than a little bit. If a room/zone doesn't have a peak load of a LEAST 2/3 the output of the room coil head or cassette, it should NOT have it's own unit.  With mini-duct cassettes the output can be split between adjacent or nearby rooms to better match the output to the load, and even though they are lower efficiency than a 1-off right sized mini-split, the hit in efficiency isn't as bad as what 2-3x oversizing would do to it.

Ducting heating/cooling air into closets it a bit silly, and not required by code.

Oddly, even the Mitsubishi rep who came down was talking about using seven heads in the house. This house is going to have open cell foam in the 2x6 walls, plus one whole side is SCIP panels. It won't be code minimum R value. My hunch is the job could be done with 4-5 heads and 3.5 tons. Can anyone refer me to a bang-up, honest hvac contractor experienced with mini splits in the Atlanta area?

Really, I think I need a recirculating air system, to be sure all of the house has air movement. And then, about four heads.

I think I need a recirculating air system

You need an ERV or an HRV. They constantly resupply the home with fresh air, reclaiming energy as it exhausts the stale air.

If you direct the linesets into an interior wall, be forewarned that the installer didn't do too good of a job "rounding" the corner and there were issues with sheetrock and corners.