Every load calculation begins with two numbers taken from a published table of local climate data, derived from roughly thirty years of hourly observations.
The two numbers
The 1% cooling design dry-bulb is the outdoor temperature exceeded in only 1% of hours during the cooling season — about 35 hours a year.
The 99% heating design dry-bulb is the temperature that outdoor conditions stay above 99% of the time in winter.
There is a third that matters in humid places: the coincident wet-bulb, which gives the humidity present when the dry-bulb is at its design value, and which sets the latent portion of the load.
Why not the record
A system sized for the hottest hour ever recorded would be oversized for essentially the entire year.
The consequence is not merely wasted money. It is short cycling, poor dehumidification and uneven rooms for 8,750 hours a year, in order to handle conditions that occur for a few hours in a decade.
The 1% point is a deliberate trade: the system may fall slightly behind on a handful of extreme afternoons, and in exchange it operates correctly the rest of the time.
What "falling behind" means
Not failure. On a design-plus-five-degrees day, a correctly sized system runs continuously and may hold the house two or three degrees above setpoint through the hottest part of the afternoon, recovering in the evening.
That is the intended behaviour. A system that never runs continuously has capacity it does not need.
Where to find the numbers
ASHRAE publishes design conditions by weather station in its Fundamentals handbook, and ACCA Manual J includes tables for US locations. Load calculation software carries the data and selects by location.
They are station-specific rather than city-wide, and within a large metro area the difference between an inland station and a coastal one can be several degrees plus a meaningful humidity difference.
The heating side and the heat pump problem
For a furnace, sizing to the 99% heating design temperature is straightforward.
For a heat pump it is not, because capacity falls as the outdoor temperature drops — exactly when demand rises. A heat pump sized to meet the full load at 5°F is enormously oversized for cooling in summer.
The usual resolutions are a balance point with supplementary heat below it, cold-climate equipment that holds much more of its capacity at low temperature, or a dual-fuel arrangement pairing the heat pump with a gas furnace for the coldest hours.
Design conditions are not static
Published design temperatures are revised as the underlying climate record is updated, and in many locations the cooling design values have moved upward over successive editions.
Sizing software using an old dataset will use an old number. It is worth asking which edition a calculation used, particularly in regions where summers have measurably lengthened.
What this means for a quote
A contractor should be able to state the design conditions used. It takes one line.
If the answer is a temperature far above the published 1% value for the area, the load has been inflated at the first step, and everything downstream of it — equipment size, duct sizing, cost — is inflated with it.
That single question is one of the quickest ways to tell whether a real calculation was performed.
The indoor half of the design condition
The other end of the temperature difference is the indoor setpoint the calculation assumes, conventionally 75°F in summer and 70°F in winter.
A household that keeps the house at 70°F in August has a larger load than the standard assumption, and it is worth saying so before the calculation is run rather than discovering the gap in the first heatwave.
Indoor design humidity matters too. Manual J commonly assumes 50% relative humidity indoors; a household that wants 45% is asking for more latent capacity and longer run times.
Elevation and the correction nobody mentions
Air is less dense at altitude, so a given volume of air carries less heat.
At Denver's elevation the correction is around 15%, and it applies to both equipment capacity and duct sizing. Load software applies it automatically when given a location; a hand calculation using sea-level figures will be wrong in the direction of undersizing.