Why the new system should probably be smaller than the old one
Matching the tonnage of the unit you are replacing is the most common and most expensive mistake in this trade.
The fastest way to quote a replacement is to read the model number off the old condenser and fit the same tonnage. It takes two minutes, it is what most companies do, and it is wrong more often than it is right.
Why the old size is not evidence
The old unit was sized by somebody using the same shortcut, one system earlier. Meanwhile the house changed. Attic insulation was topped up, windows were replaced, the west-facing slider got a shade. Each of those lowers the cooling load, and none of them changed the badge on the condenser.
What oversizing actually does
An oversized system cools the air to the thermostat setting quickly and then shuts off. That sounds efficient and is the opposite.
- It short-cycles. Starting is when a compressor wears. More starts, shorter life.
- It does not dehumidify. Moisture comes off the coil over time, and a system that runs for eight minutes never gets there. The house reads cold and clammy.
- It leaves the house uneven. Short runs never move enough air to reach the far bedroom.
- It costs more to buy and more to run.
What a load calculation involves
A Manual J calculation measures the house rather than the equipment: floor area by room, orientation, window type and area, insulation levels, ceiling height, infiltration and how many people live there. It takes about an hour on site.
In this valley the answer commonly comes back half a ton to a full ton below the outgoing system. On a recent Kingsburg job the existing unit was four tons and the calculation said three.
The questions worth asking any contractor
- Will you do a Manual J, and can I see it?
- What was the calculated load, and what size are you quoting?
- If those differ, why?
- What is the design temperature you used?
A contractor who cannot answer the second question is guessing. That is not always a disaster, but you should know that you are paying for a guess.