Most PM intervals are inherited, not chosen — copied from an OEM manual or from whoever set the system up. This finds the interval where total expected cost is actually lowest.
Most maintenance intervals get argued about using the wrong number. People reach for mean time between failures, but MTBF tells you how often a thing breaks, not how much warning you get — and warning is what an inspection is buying.
The right number is the P-F interval. Failures rarely happen instantly. There is a point P where the problem first becomes detectable — a bearing starts running warm, oil analysis shows metal, a reading drifts — and a later point F where the thing actually fails. The gap between them is your warning window.
That gives a blunt rule: if you inspect less often than the P-F interval, you will miss failures no matter how good the inspection is. A perfect quarterly inspection cannot catch a problem whose entire warning window is three weeks.
For each candidate interval it adds two costs over your planning horizon:
The optimum is where the sum is lowest. Below the P-F interval you are paying for inspections that cannot catch anything extra; above it, you start missing failures in proportion to how far past it you go.
Vibration analysis on a bearing might give months of warning; oil analysis, weeks to months; audible noise, days. The same bearing has a different P-F interval for each detection method — which is why upgrading the inspection technique often beats shortening the interval. Many reliability programmes deliberately inspect at half the P-F interval to leave room to plan the repair rather than scramble.
This assumes a roughly constant failure rate, which is a real simplification — many components have a wear-out curve where risk climbs steeply with age, and a few have infant mortality where a fresh intervention makes things worse before better. It also assumes inspections are independent and that you know the P-F interval, which is usually an estimate. Use this to sanity-check an inherited interval and frame the conversation, not as a substitute for reliability analysis on a critical asset.
The strongest input to a PM interval is not a model — it is your own completion and failure history. Once you have a few cycles logged against an asset, the data tells you directly whether you are over- or under-maintaining it.
That is the argument for capturing measurements rather than just completions. A PM record that says "done" tells you nothing. One that says the screw clearance was 0.9mm, then 1.2mm, then 1.6mm tells you exactly when the next one is due — see the screw and barrel wear inspection for what that looks like in practice.
Pick your industry and ChatterFix builds the equipment hierarchy, PM schedules, quality checklists, and safety inspections to match — in about three minutes.