On my old house, I installed a nice big PV system... and then installed tons of whole-house A/C, which used twice as much net power as the solar panels generated. Whoops. On my new house, I'd like to actually achieve net zero electricity use every month, instead of just hoping the panels will be big enough.

The new house is a 100-year-old Craftsman in Los Angeles, with 4500 square feet, lots of windows, wide eaves, no A/C, no insulation, and big gable vents in an unfinished attic. It's nice and cool most of the time downstairs... but during heat waves, the asphalt shingle roof reaches 178 degrees, the attic is well over 100, and our upstairs office reaches 89 degrees on the floor and 92 degrees on the ceiling. We anticipate having to cool about half the days of the summer.

The house has room for about 18 solar panels on the roof, so after installing PV, my daily power budget will be about 25KWH (or 30KWH if I spring for 300W panels). Our power usage was originally about 40KWH/day. As a first step towards net zero, I've already converted the whole house to LEDs, set computers to sleep when not in use, and replaced an old fridge, and we now use 20KWH/day -- which can be comfortably supplied by PV panels, leaving about 5 or 10 KWH/day for A/C. Hah!

So, first order of business is to reduce the amount of heat getting into the house. Measuring temperatures with an IR gun in the office on a hot day, I see the walls are cool, but the ceiling and window blinds are hot (5 and 2 degrees hotter than the floor at 4pm, respectively). The windows are double-paned, so I suspect most of the heat is coming in by conduction from the attic, and some as light through the windows. A quick test with one roll of R-38 in the attic lowered the temperature of the ceiling under the insulated area by 3-4 degrees. I expect heat-reducing film on the windows will also help a bit. We need to reroof at some point, and putting on Energy Star shingles might help a tiny bit, too.

But even with all that, I expect we're going to need some A/C in the master bedroom (400 sq ft) + attached office (250 sq ft), and in the kids' bedroom (300 sq ft).

Since I have an insanely low amount of power available, I looked for the most efficient minisplit around; seems to be 30 SEER, 9000 BTU (so 300 watts). With my 5KWH budget, I could run two of those flat out for 8 hours/day.

This being a historic craftsman, my wife doesn't like the ugly factor of the wall units. Ceiling cassettes might be prettier, but they probably mean cutting joists (ouch) and would probably be less efficient.

So after all that setup, here are the questions:

• Are high-SEER cassettes available, and do they work well?
• Know of any historic Craftsman houses with minisplits installed that didn't end up being ugly?
• Has anyone else tried cooling just two or three rooms of a big, drafty old house like this?

First things first.

If the wall cavities are empty, squirt them full of cellulose (any density). Even though that's a tiny fraction of the sensible load, it contributes to the infiltration losses that won't necessarily show up in an IR scan.

Air seal the upper floor ceilings, then bring the insulation levels up to at least 8" anywhere you can. If you can't get that much insulation in there put in as much as you can, and staple vapor permeable radiant barrier (the perforated aluminized fabric type, and definitely NOT aluminized bubble pack!) on the under side of the rafters, even on north facing pitches, or if it's in the budget, insulate the underside of the roof deck with open cell foam. (There are lots of particulars to that last approach before launching into it.)

Most of the heat gain is through windows, particularly bad are WEST facing windows, since they aren't shaded by the overhangs when the sun is low in the sky, and the gains occur late in the day when the roof & siding are already hot. Exterior roll down shades can cut those west-window gains by 2/3 or more. Most south facing windows on bungalows are in full shade at mid-day, and much less of gain problem than west facing windows. East facing windows are high gain and may be worth treating, but at least those gains are earlier in the day, and don't contribute as much to the peak loads.

If exterior shades aren't an option, heat rejecting window films really do work for west facing windows. Putting the window film on an exterior storm window is preferable to putting it on an antique double-hung. Don't go for the max-rejection versions though, as it may darken the room too much. If all your windows are single-pane antiques, it's worth installing some tight low-E storm windows throughout. (The Larson Silver or Gold series sold through box store chains are pretty good, their Bronze series leak too much air. ) The low-E storms don't reject as much heat as window films but they do reject a good fraction, and they reduce infiltration gains/losses.

Ignore the SEER rating of the mini-splits- pay more attention to sizing them correctly for the loads. A 9000 BTU/hr head is going to be WAY oversized for the individual room loads on most arts & crafts bungalows, especially if you've taken some or all of the above load reduction measures. The Mitsubishi dedicated single zone SEZ/SUZ series mini-duct cassettes can split the load to as many as 4 rooms through short duct runs, but it's important to keep the cassette and ducts completely inside of your air-sealed and insulated boundary. The SEZ-KD09NA may only rate an SEER of 15, but by sizing it correctly it will be sipping power compared to a 9000K head that's 4x oversized for the single-room load, and have a higher as-used efficiency.

http://usa.mylinkdrive.com/uploads/documents/4297/document/17_8_SEZ_Ducted_Heat_Pump_Systems.pdf

It's unlikely that the whole house load will exceed 3 tons if you air seal & insulate it, and use some judicious load reduction on unshaded west facing windows.

Pictures (or a google street view) can sometimes be useful in coming up with cooling load reduction strategies.

Thanks. I've attached pictures of the east and south sides of the house. The eaves protect the east and west sides nicely in mid-day, but not when the sun is low. The south side is completely unprotected, so a fair bit of heat comes in that side, I think.

The south and west windows are single-pane. I might consider replacing them, but as it's a historic house, we'd have to match the look exactly and get city approval.
Window film (we're looking at 3M Prestige 70 or 3M Night Vision 45) seems less trouble.

A short duct instead of a ductless would probably do wonders for esthetics.
The attic is currently unconditioned, and I'm having trouble convincing myself to air-seal it. That seems so... un-craftsman-like. But I'm all for insulating the roof, and the duct runs should be really short.

Got a link for how to air-seal ceilings? Presumably you mean like http://www.buildingscience.com/documents/guides-and-manuals/gm-attic-air-sealing-guide in general?

(Also, why do all the online A/C sizing calculators say I need huge A/C's? It's like buildingscience.com says, they all give values 50% larger than you seemed to just now.)

The attachments don't appear.

Over the weekend (assuming you can be around mid-day on weekends) observe the amount of direct sun coming through the southerly windows between 10AM-2PM.  It's now several weeks after the equinox- the shading factors will be much better during the peak cooling season than they are right now, since the mid-day sun is higher then.

Single-pane west facing windows are a huge heat gain unless the low angle of the PM sun is blocked by adjacent buildings, tall hedges/trees etc.

Exterior low-E storm windows don't dramatically affect the look of the house, but historical commissions may not approve them. It's a heluvua lot cheaper to put up low-E storms than to get the equivalent performance out of even a cheaper replacement window, let alone a high performance custom "architecturally appropriate" window.

Air sealing the attic floor doesn't affect the character much, but depending on what you have that can sometimes be more work than air-sealing at the roof deck.  If the roof lines are simple enough that it's easy to put eaves-to-ridge venting you can insulate leave it as a vented roof deck using a combination of  batts (rock wool preferred, but high density fiberglass is OK, not great due to it's higher transparency to infrared radation) and rigid foam.  Do do that you would need to fit the batts carefully between rafter elements leaving a minimum of 1" of vent channel between the batt & roof deck (1.5-2" is better), then put rigid foam board on the underside of the rafters.  If you can get at least R15 batts (3.5" deep, so you'd need a minimum of 2x6 rafters), then put 2-3" of foil faced fire-rated Thermax polyiso under that you can cut the roof gains by a large fraction.  Tape the seams of the polyiso with 2" FSK tape (purpose made duct aluminum duct tape), and go over the tape with 1/8" of duct mastic, it will be pretty air tight. The tougher to seal parts would be at the ridge peak and where the foam board hits the joists- you need to put an air-dam of cut foam between the joist, sealed to the joist and top-side of the ceiling below with can-foam.  You could also use non-fire rated polyiso, but that would require installing half-inch sheet rock on the interior side as a thermal barrier against fire.  If you use the attic for storage you'd have to install the thermal barrier anyway, and you could use anybody's polyiso.

Polyiso is the preferred foam in that application, since it uses a fairly innocuous blowing agent (pentane), and the foil facers mitigate any outgassing or polyols & fire retardents by several orders of magnitude.  It also has a high ignition temperature relative to polystyrene, and even when it's on fire it chars in place rather than melting sending puddles of burning plastic, as polystyrene (EPS or XPS) do. As long as the roof deck is vented, you can use polyiso. If not, you'll have to use something else- open cell spray polyurethane against the roof deck would be the next-best option.

The Building Science Corp guide is pretty good. If you're a visual type of person you can search for videos on the topic- they're out there. This site has lots of tips on air sealing and insulation in various places too. 

The key is to fix the big leaks first. Chimney & flue chases are often huge basement-to-attic air leaks, and need special treatment- the air barriers that contact the chimney or flue have to be metal, with fire-rated foams or duct-mastic sealing the seams. Sealing them at both the top & bottom of the chase counts, since the longest free path is what defines the "stack effect" pressure drive.  Similarly, plumbing & wiring chases need to be air sealed both top & bottom (where you can) but it's OK do use cardboard/foam-board or can-foam to seal those off. Insulation that contacts chimneys & flues has to be something fireproof- wrapping them with an R15 rock wool batt held in place with steel wire is a common way of prepping an attic for insulating an attic floor with blown cellulose.  You would need to do a similar sort of thing if insulating at the roof deck- the polyiso should have at least R15 rock wool between it and the chimney or flue, but you could then trim it flush with the polyiso and use a sheet-metal air-barrier over the rock wool.

Plumbing & electrical penetrations of the attic floor add up too.  Ceiling light fixture boxes are sometimes easier to seal from the bottom, pulling the fixture and touching up any gaps or holes with fire-stop foam.  Wiring is often routed up through top plates of partition walls or exterior walls, and a shot of foam can seal those off too.  If the house is balloon framed it may have no top plates in some or all walls, which can be blocked with cut-up corrugated prior to insulating. Partition walls with no top plates can be a real PITA, if they are under attic flooring. 

If you are installing a mini-duct cassette in the top floor ceiling you'll have to build out an air-sealed enclosure over it and  insulate above that. It's sometimes easier to install the cassette between the existing joists, then run a set of 2x6 or 2x8 joists perpendicular those joists, to allow routing of 4-6" ducts. Then an OSB or plywood barrier over that will give you something to air seal and insulate over. The last thing you want to do is run ducts in a 150F attic without substantial insulation over it.  Often there will be a closet space or something where the cassette can be installed in the top of that space, but not always.

Many online calculators a pure junk, guaranteed to oversize. The better ones use ACCA Manual-J methods, which do a bit better.  Using a whole-house simulation tool like the freebie download BeOpt does it even better.  But like any calculation tool garbage in = garbage out.  Specifying the R-values is pretty easy, specifying the U-factors and solar heat gain coefficient (SHGC) of older windows with window films is more difficult, and the shading factor can make a huge difference to.  For a code-min new house with pretty-good windows and a 15% window/floor area a cheap rule of thumb like "a ton per 1000 feet" usually works, and at least won't oversize it by 2-3x.  Century old bungalows tended to have lower glazing ratios and simpler footprints, but better roof overhangs for killing off mid-day window gains.  If you upgrade the house with better attic insulation and installe exterior shades &/or heat rejecting replacement windows or window films on the sun-drenched west windows most of those grand old-girls come in closer to a ton per 1500' of conditioned space, which was the basis of my WAG load number.

An oversizing of 50% is not an efficiency & comfort disaster for a mini-split, but above that is bad for both. Overshooting by 2x or 3x is a VERY common situation, one that we are taking pains to avoid.  A lot of old-school contractors use a "ton per 500' " rule of thumb, which would be more than 2x oversized for most houses like yours post-upgrades.  Under CA Title 24 a Manual-J calculation is required for sizing HVAC equipment unless you have an even better engineering model (BeOpt is better) but it's easy to cheat Manual-J with a thumb on the scale for the input data, and many contractors still do, preferring to intentionally oversize rather than risking the call-back from a sweaty irate customer.

A sanity check on where more insulation may stop making sense lives in table 2, p10 of this document.  Los Angeles CA is in US climate zone 3. While it's unlikely that you'd be able to fit R50 on the attic floor over the majority of the area you might be able do come close. It may require running a set of perpendicular joists, and you would have to install insulation barriers to keep a 1" clearance to the roof deck, it would cut your roof gains effectively down to nil. If buying replacement windows, take heed of the recommendations for U0.30 max, SHGC 0.3 max.

Awesome info, thanks.

As the eaves on the south side are very high up, they don't shade much, and I suspect light comes in the windows even at noon. But I'll check this weekend.

The photos are attached to the original post, I couldn't figure out how to attach them to the reply. So look up, they're there, taken at about 3:30 pm yesterday. Note how the eaves shadow is just touching the top of one window, but several feet above the other.

Air-sealing the chimneys (3!), plumbing, and dumbwaiter (!) races for this house would be quite the project. Even sealing the ceilings (and, as you point out, the tops of the walls) over the three rooms I need to cool is daunting, but at least finite and not hopeless.

Our roof is amazingly old; it's asphalt shingles applied directly on top of the old wooden shake shingles, on top of rafters... the wooden shakes are clearly visible from the attic!
It needs replacing. Not sure if soffit vents are possible - it'd be a lot of hole-drilling - but when we re-roof, I will push for soffit + ridge vents and the kind of under-the-deck insulation sandwich you describe.
(Rigid foam insulation on top of the deck is tempting but might make the edges of the roof look funny, so would probably not be allowed by our HPOZ and the city's Mills Act board.)

I just figured out that you added the attachments to the original post, not the reply.  Nice house!

It looks like you have cathedral ceiling in some of the built-out dormers?

You probably get a ton of AM solar gain on the east facing upper floor through all of those windows, which kind of pre-heats the upstairs- great in winter, sucks in July.  Looks like maybe 100-120 square feet of glass (?). The bottom floor windows on that side look well shaded from the deep porch roof and the shrubs.  If the upper floor windows on the west side are similar, that has a bigger effect on the peak loads.

The south side shot looks like it was taken maybe 4-6PM when the sun is low in the sky.  When angle of the incident sun is more than 45 degrees off the plane of the window surface, a large fraction of the light & heat is reflected off the exterior, compared to how much is reflected when it's less than 30 degrees off the surface. (The amount reflected is a function of the differences in refractive index of the glass compared to the refractive index of the air, and the angle.)  The bigger gains would be during the middle of the day, if any of that south-facing glass is in direct sun. The overhangs relative to the windows look pretty good for cutting the July/August heat, but it might start picking up in September. it's hard to tell from just a picture, but the solar geometry is well defined.

Looks like you have at least two chimney chases to air seal.

Because you have stucco siding the wall stackup needs to be investigated a bit before pounding the wall cavities full of insulation. Ideally you would have plank sheathing on the studs with some tar-papery or asphalted felt layer on the exterior of that, with an air-gap between the felt and lath.  But in drier climates it was also common to just cut in bracing for structural support against racking forces, then put the lath directly on the studs.  If your stack up is the latter there is some risk to insulating right up to the lath, particularly on any location that gets rain wetting on the stucco. The stucco can hang onto lot of water (even from dew), and when the sun hits it the moisture drive toward the interior is intense. With an empty wall cavity the lath will dry quickly into the cavity, but if it's stuffed full of air-retardent insulation it slows it down a lot.  On the sunny south side it would still dry toward the exterior fast enough to not matter (much) but on other sides it may not.  An insulation contractor worth their salt would scope this out ahead of time, but unfortunately they aren't all building-scientists, and it could take years or even decades before problems related to those conditions became evident.  (There are no cheap & easy retrofit solutions if the lath is hard up against the studs, with no sheathing, tar paper, or air gap.)

Looks like we're cross-posting at the same time, eh? :-)

Since it looks like you're re-roofing, you still have options:

In US climate zone 2 you're fine with just R5 under the shingles to go unvented if you are re-roofing, which would allow you to put as much as another R33 (for a total of R38- the IRC 2012 code-min), as fiber on the interior. See:

http://publicecodes.cyberregs.com/i...sec006.htm

http://publicecodes.cyberregs.com/i...sec002.htm

If you strip even the shakes off you can air-seal the roof deck with a self-healing membrane such as Grace Ice & Water Shield and put down a 1" layer of foil faced polyiso, (which wouldn't be thicker than the shakes) or 1.5" of EPS, with a layer of #15 felt under the new shingles, which can be long-nailed through the foam to the roof deck.  If you can get away with 3" foam it would be better, but it would require a half-inch OSB or plywood nailer deck through-screwed to the rafters at least 24" o.c. for attaching the felt/shingle lay-up.

With a 1/4- 3/4" adder to the roof stackup, it's unlikely that the historical commission is going to flunk you on that, but they might balk at 2.5-3", which is where you would be with 3" foam even if you strip it down to the roof deck.

Under CA Title 24 you'll probably need a "cool roof" shingle anyway, but be aware that the shingle temps would otherwise run higher temps when you insulate the roof deck.  On roofs with your amount of pitch that's not a huge concern- it'll convection-cool on the exterior, but on 2:12 or lower you would want to use a fairly high solar reflective index (SRI) shingle to keep shingle temps down.  The hit on shingle life is still less than 10% even if it runs hotter, but it's worth keeping in mind. On your roof an SRI of 30 or higher is probably Title-24 legal, and will do just fine. (You can get SRI 30+ shingles that appear even darker than what you currently have.)

Yeah, I can strip off the old shakes.

Can you recommend SRI 30+ shingles that look darkish? I know about Landmark Solaris asphalt shingles and Inspire 'slate' shakes, but I don't think Home Depot carries any of those :-)

I'm willing to insulate under the roof, and the 1" polyiso layer on top sounds doable, but how can I justify it? What's the payoff over just insulating under the roof -- is it to reduce heat bridging at the joists?

I'm attaching pictures of the attic overall (note the existing furnace ducts going hither and yon) and of a soffit from the attic side (which shows the back-end of those cosmetic beams visible from the outside, and gives some hope one could drill soffit vent holes).

Thanks again for the advice about wall insulation and moisture. Yes, our walls probably were built without any of the modern techniques like tarpaper or felt. They're pretty basic.

We are having a heck of a time getting our contractor to get moving on this project. I'm probably going to throw my hands up
and let him use blown-in insulation in the attic. (I hate the thought of not being able to see the wood, and dread the thought of him accidentally trying to fill the old dumbwaiter shaft or an interior wall, but everyone says batts are just too hard to do right.)

For shingles, my wife vetoed anything but the most minimally cool roof, so we're using composite shingles which just barely meet the city's requirement, Certainteed Landmark Solaris Dusky Clay, product ID #0668-0056.

Biggest lesson: call the contractor every few days no matter what...

Dan, Or... if he is blowing you off: get a new one. Call every evening (that's when he is making his plan for tomorrow).
Good luck!
richm

Finally got his attention, have a tentative date to start work, all is well (knock on wood)

Wish I'd been way more proactive about making sure the ball was always in his court, and that he heard from me every day or two, though. Would have made for much less nail-biting.

Update: we did get the new roof and solar. All that is working well, see
http://kegel.com/energy/solar/retrofit2/
We did nothing special about venting, just kept the existing large vents
But aside from rolling out one roll of batting and noticing that, yup, it did change the temperature of the ceiling just under it as expected, I haven't done diddly.

The summers have been brutal, and we're thinking harder about installing A/C.

My goals:
- can keep 2 rooms cool on solar power alone from 11am to 3pm, when we have >= 3kW of solar to spare
- reduces our carbon footprint (so must be a heat pump that can replace our upstairs furnace, too)

My wife's requirements:
- cool five rooms upstairs
- no ugly non-historic-looking inside units

City requirements:
- no concrete slab (it's a historic landmark home with a Mills Act contract)
- setback issues mean outdoor unit must be very slim and/or fit on roof somewhere unobtrusive

House constraints:
- 16" spacing from start of one rafter to start of next
- Dimensional lumber, so there's a 14" gap between rafters
- Hand-measured, so gap between rafters varies +- 3/4"

The five rooms I'm trying to cool are maybe 2000 square feet total, and I'll only be cooling
two thirds of that usually, so 1500 square feet.... which you suggested could be handled by one ton / 12,000 BTU.
But it's drafty and poorly insulated, only a few of the many windows are double-paned,
and I'm probably only going to be able to insulate the attic, not the walls (which are mostly windows anyway).

Happily, minisplit choices seem to have gotten better since I first posted.

Mitsubishi has multihead systems with units that fit between 16" joists (model MLZ).
Unhappily, the smallest MLZ I can find is 9000 BTUs... and the smallest compatible
outdoor unit that is listed as being able to drive five of them is the 40,000 BTU cooling
MXZ 5C42NA2. That seems wildly overpowered by your remarks above,
but maybe given just how leaky the house is, maybe that's not so far off.
The MLZ is designed for modern homes, so when it says it fits a 16" joist
spacing, it wasn't thinking about old homes with dimensional lumber;
the spec sheet says the hole has to be 14 13/16". I have to run around
in the attic and see if there are any good spots with that much spacing.
If not, maybe I'll have to consider some other option. (Or maybe
we could sister the joist and shave off 13/16" of wood, if that's not too dumb.)
Two of the larger rooms I'll be actively cooling could well share
one short-ducted in-ceiling indoor unit; the others probably all want their own zones.

Oh, and while we're fiddling with the electrical panel (it doesn't have space
for A/C), I'll probably add breakers for an EV charger and future
electrification of the water heater and downstairs HVAC. And maybe
install a battery, depending on how much money I want to spend...

Sigh. Now I guess I need to start looking for an energy retrofit contractor again.
I can't imagine there are many, even in LA, that have experience
doing in-ceiling minisplit heat pumps in historic homes and
know how to calculate real loads.

Have you run a room by room Manual-J on this house (particularly the rooms that are "must cool"), on the "after attic insulation and other upgrades" version of the house?

Even a freebie online version such as LoadCalc is way better than a WAG. Be aggressive rather than conservative on things like R-values and air tightness. Even when assuming the house is hermetically sealed it still overshoots reality by a double digit perctenage on my (insulated but sub-code) 1920s bungalow.

https://loadcalc.net/

A step up might be (but isn't always) is CoolCalc, which is an ACCA listed Manual-J tool. But like any tool, understanding how to use it correctly is critical to getting accurate results.

https://www.coolcalc.com/


When married to a dedicated SUZ-KA09NA2 compressor the MLZ-KP09NA can modulate down to 3600 BTU/hr @ 95F outdoors / 80F indoors, which may still be pretty high for a single bedroom:

http://meus1.mylinkdrive.com/files/M_MLZ-KP09NA_SUZ-KA09NA2_SUBMITTAL-en.pdf

When hanging on an MXZ multisplit it's a nominal 9000 BTU/hr, and any modulation range is limited by the minimum modulated output of the MXZ and what other loads it's serving:

http://meus1.mylinkdrive.com/files/MLZ-KP09NA_For_MXZ_MULTI-ZONE_SYSTEMS_Submittal-en.pdf

The minimum modulation level of the -5C42NA2 is 12,600 BTU/hr @ 95F out, 80F in:

http://meus1.mylinkdrive.com/files/MXZ-5C42NA2_Submittal.pdf

Unless the loads of the 5 rooms pretty much rise & fall together these may end up cycling the compressor on/off a lot with a minimum output that large, which takes a bite out of the potential efficiency. When running any single zone the compressor has to cycle on/off no matter what, since it's minimum output is higher than a single MLZ-KP09NA can handle.