Inspection frequency in condition-based maintenance is anchored to a single engineering rule: the check interval must be less than half of the P-F interval for the failure mode you are guarding against [S1][S2][S8].
The P-F interval is the elapsed time between a detectable potential failure (P) and a functional failure (F), and it can range from weeks for a fast-degrading bearing race to six to twelve months for a gearbox picked up by wear-particle trends in oil analysis [S1][S3].
The Half-Interval Rule and Where It Comes From
The "half of P-F" rule of thumb exists to absorb two real-world uncertainties: the exact moment a fault crosses the P threshold, and the scatter in the P-F duration from one failure instance to the next [S1][S2][S8]. If a bearing outer-race defect gives a 6-week P-F interval, scheduling condition monitoring every 3 weeks means a worst-case missed reading still leaves about one interval of warning before functional failure, which is the cushion the rule is designed to provide [S8].
The concept is rooted in Reliability-Centered Maintenance (RCM), formalised in a 1978 Nowlan and Heap report for United Airlines and popularised by John Moubray's "RCM II" framework; the P-F curve is a planning tool, not a prediction model, and it shows the action window once a failure has already started rather than forecasting its onset [S3]. The interval is a property of the failure mode and the chosen detection technology, not of the calendar [S3][S4].
Why the P Point Shifts With Detection Technique
The same physical fault produces different P-F intervals depending on what instrument is listening for it, which is why the chosen technique has to be matched to the failure mode before the interval is locked in [S1][S3]. Ultrasound detection of friction, impact, and turbulence is among the earliest techniques to flag a developing bearing fault, often placing P further left on the curve than vibration or temperature [S1]. Vibration analysis remains one of the most widely used P-point indicators for rotating equipment, tracking bearing defect frequencies, imbalance, misalignment, and looseness [S1].
Oil analysis (wear-particle concentration, viscosity shift, contamination) tends to flag internal gearbox and hydraulic degradation months before audible or thermal signs appear, while infrared thermography and motor current signature analysis catch electrical resistance, insulation breakdown, and rotor-bar faults [S1][S3]. Performance indicators such as flow rate, pressure drop, and energy consumption sit late on the curve and leave the shortest usable P-F interval, so they generally require shorter check intervals than condition-based techniques [S1].
Decision Matrix: Matching Interval, Technique, and Criticality
Different failure modes support different monitoring cadences, and the half-of-P-F rule interacts with both criticality and the chosen detection technology to set the final inspection frequency. The comparison below lines the main options up against three decision criteria: how short the P-F interval typically is, which technique places P earliest, and the criticality tier that justifies the cost. [S2]
Short-interval failure modes (fast-bearing spall, short-circuit onset, thermal runaway) pair with vibration, ultrasound, or infrared at intervals measured in days to a few weeks, and they are typically reserved for safety-critical or single-train assets where a functional failure stops the line [S2][S5]. Long-interval failure modes (gear wear, lubricant degradation, winding insulation ageing) pair with oil analysis or partial-discharge trending at intervals of weeks to months, and they suit lower-criticality assets where the cost of continuous monitoring is hard to justify [S1][S3]. Highly critical production assets that directly affect throughput, worker safety, or product quality should be prioritised over equipment with minimal operational impact, with a criticality assessment used to allocate monitoring resources before the half-of-P-F rule is applied [S2][S5].
Continuous online monitoring is justified when the P-F interval is short, the failure consequence is severe, or the asset is inaccessible during operation, because a fixed calendar interval cannot catch faults that develop faster than the gap between readings [S4]. Route-based portable checks remain the cost-effective default for non-critical or redundant equipment, where doubling the inspection headcount to halve the interval does not pay back in prevented failures [S2][S4].
Limits of the Rule and Common Failure Modes
If the P-F interval is genuinely short, even a 90-day route interval can let a fault cross F before the next reading, which is the failure mode the half-of-P-F rule is explicitly designed to prevent [S6]. The rule assumes a single dominant failure mode with a known, roughly repeatable P-F duration; when several failure modes compete on one asset, the shortest practical P-F interval should drive the cadence, and additional techniques may be layered on top rather than run on the same calendar [S1][S7].
The 50% margin also assumes a detection technique that has been calibrated to the failure mode, and a misapplied sensor that surfaces the fault late will silently shrink the usable P-F window without changing the rule [S3]. Estimating the P-F interval itself is judgement-based: oil analysis, for example, can give six to twelve months of warning on a degrading gear set, while an infrared camera on the same gearbox may only flag elevated temperature a few weeks out, and the same asset can therefore need very different intervals for different failure modes [S3]. The rule is therefore not a substitute for a failure-mode-and-effects analysis; it is the scheduling step that runs after one.
What a Trackable Program Looks Like
A defensible program starts by listing each asset's failure modes, estimating the P-F interval for each, and selecting the technique that places P earliest, then scheduling inspections at no more than half of the shortest P-F interval in that list [S1][S2][S7]. A practical first pass is to log the assumed P-F interval alongside the chosen technique in the CMMS so that intervals can be re-tightened as real failure data accumulates.
Two signals worth tracking over the next planning cycle: the share of work orders generated by condition findings versus calendar triggers (a mature program should sit predominantly on the condition side), and any asset where the recorded P-F interval turns out shorter than the original estimate, which forces a tightening of the check cadence on the next route revision. The mechanics of how vibration-based condition monitoring and broader condition monitoring systems feed those work orders are covered in the linked references.
Detailed specification references: power monitoring system.
Related analysis: Cold Box Core Vent to Blow Hole Area Ratio: Spec and Defect Guide.