I have 3 quotes for HVAC systems for my house. 2 are for 4-ton systems and one is for a 5-ton system. From what I can tell these are all from contractor rules-of-thumb (X tons per Y sq. ft) and contractor rules of thumb are notoriously oversized.

But I don't want an undersized system. My current 1800 square foot house has a 3-ton system that strains in the summer. My new house should be much better insulated, with blown blanket insulation, white roof, lattitude-orientation and a tighter thermal envelope, but more square footage and higher ceilings in some places as well.

And to top it all off, I don't have mad money to spend on this. It's nearing the end of the project and it's far over budget -- I picked a "bargain" contractor who wasn't familiar with green building, and so far I've gone about 40% over budget and not only am I not going to be able to install the solar panels and landscaping I wanted, now I'm looking at just running out of money period.

Any advice on money-saving ways to get a right-sized, non-leaky AC system without paying a huge "green" markup? Should I just bite the bullet and get an independent formula J calculation done?

But I don't want an undersized system.

No one does. It's one of the reasons why the HVAC boys find it easy to upsell.

Should I just bite the bullet and get an independent formula J calculation done?

Yes, if you can't get the HVAC contractors to do a proper one you can trust.

Well I've asked the HVAC contractors so far for their Manual J's, and nothing radical seems to be there. One of them thought I had a fireplace when I don't. There are numbers I can't really understand though. It strikes me as a little bit suspicious that they put R-values for walls and ceiling when I hadn't given them any numbers from my insulation guy. (Also, the insulation guy doesn't seem to know the thickness of all the walls which means that will probably go up in budget as well...)

Thanks for the thought!

If these are tracking the 400-500 feet/ton old-school rule of thumb they're grossly oversized numbers for air conditioning unless you have a wall of west-facing windows or uninsulated leaky ducts in the attic something raising the cooling load. Even in the southern US a tight house with a heat-rejecting roofing and tight ducts inside the thermal boundary of the building you should be able to hit 1000'/ton type ratios without much problem. Rules of thumb aren't inherently bad, unless you're using an absolutely WRONG rule of thumb (as is likely in this case.) Oversizing the beast 2x would allow you to turn it off, then come home to a 90F house and still be able to cool it down in only an hour, but it'll be noisier and it'll short-cycle on you at low cooling loads.

The surest way to a highly efficient non-leaky AC system is to go ductless (mini-split/multi-split.) A 2-3 ton 3-4 head inverter drive multi-split isn't super cheap, but it's pretty flexible, and you can get them with SEER > 20 ( and HSPF > 9, if you're heating with the same unit.) The notion that you need individual heating/cooling output in every room is a fallacy when it comes to "much better insulated" homes. Figure out which space(s) have the higher load(s), and place the ductless heads there. No ducts = no duct losses, and being fully-modulating units (that get better than name-plate rating efficiency at the lower speeds) they're quieter and more comfortable too. If you take a "set and forget" approach to temperature it'll usually use less power than letting the house heat up all day an cranking it on when you get home. It pumps out more total heat to be able to maintain the lower temp, but it'll be pumping it at ~2-3x the average efficiency of doing a recovery from a big overshoot at maximum compressor & blower speed.

R-values for walls are almost inconsequential to the cooling loads, since over 90% from windows (unless you have VERY high performance windows with U-factors under 0.2 ),rather than heat conducting through the wall. R-values of attics count, but the solar reflective index (SRI) of the roofing also counts BIG TIME, since that determines the peak temps of the roof deck & attic that you're insulating against. Roof overhangs & awnings can substantially reduce gains from south facing windows, but not from east & west facing windows. West-facing glass has the highest effect on the PEAK load since it's getting the gain during the hotter part of the day, and after the sun has heated up the mass of the roof & walls. East facing glass takes on as much or more BTUs per square foot, but it takes them in when heat gains through the attic and walls are comparatively low.

Exterior shades on west facing glass can make a real difference in the peak load, but aren't included in a Manual-J, which doesn't take in that level of detail. A DOE2 or BeOpt model will give higher accuracy than a Manual-J, but it's not clear if taking it that far is necessary just for sizing the mechancals unless you're going for Net Zero Energy or some other high-performance standard for the building.

If you go with a standard ducted split system, run all the ducts fully inside the insulation & pressure boundary of the house, even if it means you have to run taller framing to accomodate a horizontal duct-chase/vault. Putting it in the attic can nearly double the peak load, and it makes air sealing (of both ducts and ceiling) more critical while complexity to the air sealing effort.

See also: http://www.greenbuildingadvisor.com/blogs/dept/musings/calculating-cooling-loads

Wow, Dana, thanks for all that info. I would love to have a net-zero home and but unfortunately in order to do that I'd have to give up some things I really like ( like eating :) )

Posted By Dana1 on 15 Jun 2012 02:47 PM
Even in the southern US a tight house with a heat-rejecting roofing and tight ducts inside the thermal boundary of the building you should be able to hit 1000'/ton type ratios without much problem.

I think you're right... the existing calculation gives 720'/ton, but it is less than 1000 BTUh from being over 900/ton... and it looks like the calculation assumes I have a chimney leaking air, which I do not and which probably would make that 1000 BTUh difference.

Exterior shades on west facing glass can make a real difference in the peak load, but aren't included in a Manual-J, which doesn't take in that level of detail. A DOE2 or BeOpt model will give higher accuracy than a Manual-J, but it's not clear if taking it that far is necessary just for sizing the mechancals unless you're going for Net Zero Energy or some other high-performance standard for the building.

I only have a few small windows on the west side, and they have deciduous trees to the west, which should help make some natural shade in the summer but the fallen-leaves should let the sun through in the winter (hopefully... this is one of the more "clever" features of the passive part of our design and I'll be giddy with joy if it proves to actually work that way.)

Another thing I noticed from the calculation, is that the heating load is higher than the cooling load. It hadn't occurred to me that in the Southwest with 100+ degree summers that I would be concerned about heating load, but I remember someone from the power company warning that it was possible. Do those alternate calculations account for passive solar features like my big south-facing windows that let the winter sun shine onto a big thermal mass of concrete?

See also: http://www.greenbuildingadvisor.com/blogs/dept/musings/calculating-cooling-loads

Nice link. I also enjoyed this essay which was linked from that post. Thanks again!

DOE2 and BeOpt do correct for passive gains, site shading & orientation, and thermal mass issues, but they're not as good as some proprietary passive solar design tools. They are US Department of Energy tools that are downloadable for free. BeOpt runs DOE2 underneath, so the results on a particular design would be similar. BeOpt was intended as a tool for tweaking different building components to find the optimal bang-for-buck on building envelope upgrades. It's been around for a decade, but was released for public consumption less than a year ago. (The NREL download site for it seems to have been off line for several weeks now, I hope that doesn't mean it's being clawed back...)

What is your location/zip code, and what outside design temps were your contractors using in their calculations? It's pretty common for contractors to use a design temp closer to the all time record highs & lows rather than the 99% & 1% numbers on a recent 25 year data set which bumps the calc considerably.

There's also margin built into Manual-J and I=B=R type calculations on the heating loads, and down-sizing by 15% would be reasonable even in a low-gain low-mass house. But in high-mass passive solar situation the mass alone takes a good chunk off the peak loads, and your true peak heating loads could easily be 25-35% lower than a Manual-J.

Also, what are your heating fuel options/rates and what are your electricity rates?

The zip is 78621, and the outside design temps look pretty reasonable, 100 in the summer (when record highs are in the 105-106 range, and last summer we had a string of 40 consecutive triple-digit days) and 30 in the winter.



The only heating fuel option we have out where we are is propane delivered by a truck. To do so requires a lot of extra expense in plumbing work, tank installation etc. that would greatly increase the cost of the build. I guess we could build a rocket stove and burn some of the abundant mesquite, too

In Elgin TX in a high performance house you could both heat and cool with ductless mini-split/multi-split. With a high HSPF (9.0+) ductless you'd meet or beat the performance of a ground source heat pump, and it would cost you a fraction of what it costs to heat with propane. (Not that mesquite in a rocket stove would be bad, but any bulk fuel heating gets to be a real PITA after awhile.)

The key to point-source heating & cooling with ductless is better-performance windows and placing the interior head(s) in the spaces with the biggest load(s). A 2.5- 3 ton 3-4 head mult-split can handle a lot of configurations.

The mean January temp in Elgin is ~48F, a temperature at which a ductless is going to running a coefficient of performance above 4. When is 30F or so it'll only be getting a COP of ~3, but at ~15F it'll still be running ~ 2.5. Your winter average will be over 3.5, but for the sake of argument, let's say it's 3.5, and say electricity goes up to 15 cents/kwh (it's currently about 11-12 cents for most of TX).

1 kwh ==3412BTU, so with a COP of 3.5 each kwh puts (3.5 x 3412=) 11,942 BTU of heat into the house. A million BTU (MMBTU) then takes (1,000,000/11942=) 84 kwh, which costs (84 x $0.15=) $12.60

In TX propane averaged about $2.80/gallon this past heating season, and each gallon contains ~91,000BTU. Say you burned it in a 95% AFUE furnace (and had ducts with zero leakage, completely inside of conditioned space), that would deliver (0.95 x 91,000= ) 86,450 BTU of heat into the house. To get 1MMBTU into the house would take (1,000,000/86450=) 11.57 gallons, at a cost of (11.57 x 2.80=) $32.40...

...which is more than 2.5x the cost of heating with ductless air source heat pumps.

If you have reasonable-R walls and better than code windows and the place is pretty tight (or could be made pretty tight) this can be a very reasonable option, but expect pushback from the HVAC crowd who still believe that every room needs a heating & cooling duct no matter how high-performance the building envelope is. The lower the overall heating & cooling loads, the lower the room-to-room differences are, and even if you have to overheat some of spaces slightly to keep everybody cozy when it drops to 20F or lower on the 99.9% decade-extremes, it's not any different from heating the main area with a woodstove- most people can figure out a way to be comfortable- it' doesn't take room-by-room micro-zoning or even room-by-room ducting when you have a decent-performance house.

See:

http://blog.energysmiths.com/2012/03/winter-temperature-data.html

http://blog.energysmiths.com/2011/12/living-with-point-source-heat.html

Posted By Dana1 on 15 Jun 2012 06:00 PM
The NREL download site for it seems to have been off line for several weeks now, I hope that doesn't mean it's being clawed back...)

Not clawed, back, but rather a new release. The BeOpt site is back up, with an updated version of the tool:


http://beopt.nrel.gov/