As others said, the design temp is not as cold as the year's low. On top of that, you don't want the geo part to be able to produce all of the heat at design temp - it makes economic sense to use electric resistance heat (or propane) for some of it. You might even find that 3 tons of geo works well - I recommend that you get someone to run all the numbers for you.

Posted By engineer on 21 Dec 2012 10:35 PM
Systems are normally designed to meet load 99% of the time via a manual J calculation. Stated another way, the system will be slightly undersized for 40-50 hours each winter or each summer. The 40-50 hours won't happen continuously, but rather in multiple 2-4 hour intervals (such as from 2 am to 6 am during a cold snap or from 2 pm to 6 pm during a heat wave). A combination of Manual J safety factors and building thermal inertia carries through the short intervals.

Design temps are published for hundreds of cities - use the one nearest your home. That said, don't be afraid to adjust for micro climate issues, such as being 2000 feet above the nearest city for which design data is published. THAT SAID, don't take carte Blanche to arbitrarily add or subtract 5-10 degrees from a reasonable design temperature - that is poor practice leading to oversizing and lost efficiency.

A rough rule of thumb that works in general (with many counterexamples to prove the rule) is to subtract ~3F per 1000' of altitude above the nearby location with a published design temperature.  One can also use resources like weatherspark.com to compare cold-day lows between your location and that of the nearby published design temp location to fine tune it a bit.

ACCA Manual-J type heat load calculations have some amount of built-in margin. Even undersizing by 10% from a Manual-J would only VERY rarely result in a cold house, even when the outdoor temps are well below the design temp. Upsizing by more than 15% from a Manual-J calc can lead to lower efficiency and lower comfort.  While people tend to worry more about undersizing than oversizing, the bigger (and very common) real-world problem is the latter.  In CA under Title 24 oversizing by more than 15% simply isn't allowed, but it still happens fairly often via errors or manipulation of the input data (intentional or otherwise mis-statement the true R-values/U-factors or design temperatures.)   When the incremental sizing of the equipment is on the edge, there is a bit of a bias by contractors & customers alike to tip to the high side, "just to be sure". But the impulse to do that is just wrong.

Agreed.

A recent client with a 2500 SF house in north Florida came to us with $300-$400 electricity bills. We found 4.5 tons of HVAC spread across two systems and SIX zones. Now, know that I like zoning, but six is nuts for that house...only two people live in it. There was a 17 year old single stage system (SEER 10) and a 6 year old two stage system (SEER 16). They inherited this situation when they bought the house two years ago, and were ready to replace all HVAC.

I played around with a load calc and determined that a well-installed spray foam job might allow us to yank the old 2.5T and reconfigure the newer 2T (and its nifty Infinity zoning) to condition the whole house via a much more reasonable 3 zones. That is certainly not the proposal the client expected from us as an HVAC contractor, but we went ahead with it. I simply was not willing to kick a modern 2 stage R410a system to the curb.

Unfortunately, the house leaks air at a much higher rate than I'd expect, and while the spray foam reduced that markedly, it is still somewhat over our benchmark of 1 CFM50 per SF living area, so we may be in for trouble next summer. Still, the client is onboard with the "experiment", and accepts that a 3 ton condenser may be needed...fortunately the existing air handler will accept that if needed.