Ask a facilities director when a given split unit is due for replacement, and the answer usually comes from a depreciation schedule — a number set at the time of purchase, based on a standard useful-life assumption, that has nothing to do with how that specific unit has actually been maintained.
It's an understandable default; the depreciation schedule is the number that's already in the system, tracked automatically, requiring no additional effort to reference. But mechanically, it is close to meaningless — a 10-year accounting assumption applied uniformly to units that might have anywhere from 6 to 20 years of real service life left in them, depending entirely on how they were maintained.
The two clocks diverge for a specific reason: depreciation schedules are set once, at purchase, using a conservative industry-standard assumption that has to hold across every possible maintenance scenario, from excellent to neglected.
Mechanical life isn't set once — it's the outcome of every service decision made over the unit's operating history, and it moves in response to those decisions in a way the accounting schedule never will.
A unit that's had:
- its refrigerant charge verified and corrected at every visit,
- coil cleanings thorough enough to restore full heat exchange,
- electrical connections checked for the resistance increases
will often run years past its depreciation schedule with no meaningful increase in failure risk. A unit that's had none of those things will sometimes fail years before its depreciation schedule runs out.
Treating the depreciation date as the mechanical end-of-life date leads to one of two costly mistakes: replacing capable equipment on a schedule that has nothing to do with its actual condition, or — more commonly — assuming a unit still has years left simply because the schedule says it should, without ever checking whether the maintenance history supports that assumption.
For a portfolio of any size, this reframing has a direct budget consequence. Instead of a wave of “scheduled” replacements arriving in whatever year the original purchase invoices happen to cluster around, replacement becomes staggered — driven by which units are actually showing end-of-life indicators, rather than which units happen to share a purchase date. That's a more level, more forecastable capital plan than one built around an accounting artifact that was never meant to predict mechanical failure in the first place.
FIELD NOTE
The next time a unit crosses its depreciation date, don't schedule its replacement — schedule a measured assessment. Let the readings, not the invoice date, decide what year it actually belongs in.