Run Hours Are the Currency an Air Conditioner Spends

Here is a piece of arithmetic worth carrying around. If a fault removes a fraction of a system's capacity, the system must run longer to move the same heat. Specifically, if it loses a fraction c of capacity, it runs 1 / (1 − c) times the hours.

Lose 20% of capacity and the machine runs 1.25 times as long. Lose 25% and it runs a third longer. The building still gets cool. Nobody calls anybody. The unit is simply spending its hours faster than it should be.

Why hours matter

Because hours appear to be what equipment life is actually measured in. The Florida Solar Energy Center's field study of 46 homes over four and a half years found operating hours were a significant predictor of shorter service life. And NREL and Purdue's fault prioritization model — the one published attempt we know of to price faults against equipment life — does it by prorating life-cycle cost against the ratio of faulted to baseline run hours:

Prolonged run time could lead to reduced unit life span and incur additional equipment and maintenance costs.

Frank, Kim, Cai & Braun — NREL / Purdue

That is a supported direction: capacity loss lengthens runtime, and runtime shortens life. It is worth being precise that the conversion inside that model is an assumption in a cost calculation — it assumes a flat ten-year life across cooling types — rather than a measurement of equipment. Which is why we will not turn it into a number of years.

Putting the two together

NIST measured a 30% undercharge at about 15% capacity loss. Run that through the arithmetic and the system is doing roughly 1.18 times its normal hours for the same cooling. Every one of those extra hours is wear it should not be doing.

The energy and the wear are not the same number, which is a subtlety worth holding on to. The same NIST test measured that undercharge at 12% lower COP, so the electricity bill rises about 14%. But the machine also draws less power per hour when it is starved. The bill and the wear diverge, and only one of them shows up on a statement.

Why this matters more across a portfolio

On one house this is an interesting fact. Across a few hundred units it is a capital-planning problem, because the effect is invisible to every system you currently use to manage the assets.

• No work order is generated. The unit is cooling. Nothing is broken enough to call about.
• Runtime alone will not rank it. A unit running long hours may be undersized, occupied differently, or sitting in the sun. Hours without a reason are not a fault signature.
• The consequence lands years later, as an early replacement, in a budget cycle nobody connects back to a charge fault four summers earlier.

And early replacements are real: of 46 monitored systems, twelve were replaced during the study at a median age of 13.5 years, against a typical service life closer to 18.

What we are careful not to claim

That a fault makes a unit wear faster while it lasts is supported. How much total life any fault costs is not — that depends entirely on how long it is left running, which is the one variable that changes when somebody is watching. There is no published figure, and we have written separately about going to look for one.

There is also a second wear channel nobody has sized. Running out of spec raises operating temperatures, and Hu, Yuill et al. expect that thermal effect to matter more than the energy penalty. It is excluded from everything above, which means the runtime arithmetic here is a floor rather than an estimate.