The P-F curve – and why Plucknette completed it.

Why one curve runs your whole PdM program

Most machine failures don't happen – they develop. A bearing doesn't explode out of nowhere: a defect initiates, then announces itself in ultrasound, then in vibration spectra, then in oil debris, then in heat, then in noise – and finally the machine stops. The P-F curve is simply that story drawn as a graph, and it answers the two questions every predictive maintenance (PdM) program lives or dies by: which technology can see the failure coming, and how often must we look?

The curve appeared in Nowlan & Heap's 1978 RCM report and was popularized by John Moubray in RCM II. Doug Plucknette – creator of RCM Blitz® and one of the most under-rated authors in the field – then spent two decades pointing out what the classic curve hides: by the time you're watching for "P", most of the die is already cast. His D-I-P-F curve (2006) completes the picture.

Three things follow directly from the picture:

• Each technology has its own P. Ultrasound and vibration see a bearing defect months before heat or noise do. "We do thermography monthly" is not a strategy until you know what failure modes thermography can actually catch, and how late on the curve it sits.
• The P-F interval is failure-mode specific – not asset specific. The same pump has a months-long P-F interval for bearing wear and a seconds-long one for a coupling shear. That's why task lists must be built per failure mode (the RCM discipline – see our RCM II summary).
• Detection buys planning time, not life. Finding the defect at P doesn't make the machine last longer – it converts a 2 a.m. breakdown into a planned, kitted, scheduled repair. The value is the difference between those two costs.

*Indicative orders of magnitude for common rotating-equipment failure modes – your machines, speeds and duty will differ. The logic, not the numbers, is the takeaway.

This is also where Plucknette's field criticism lands hardest. In The Introduction of the I-P Interval, he describes PdM programs set up as generic technology routes – monthly vibration on everything, quarterly thermography on everything – without a single inspection naming the failure mode it was looking for. The assets kept failing. A PdM task that doesn't know its failure mode and its P-F interval is a ritual, not a strategy.

And continuous monitoring changes the economics, not the logic: an IIoT sensor is just an inspection whose interval has gone to zero. It widens the usable warning window for fast-developing failures – it does nothing about defects that haven't initiated yet. For that, read on.

The proactive domain (D to P) is where Plucknette says the real return on investment lives, and what belongs there reads like a checklist for a capital project done right: RCM analysis from the design phase, FMEA, reliability block diagrams, design standards, equipment hierarchy and criticality analysis, supplier agreements, precision installation and commissioning – then the maintenance plan that the analysis produced, executed with on-condition monitoring, PM and operator care.

The distance between I (installation) and P (first detectable defect) is not fixed by fate. It is stretched or shredded by things fully under your control:

• Precision installation – laser alignment, balancing to grade, correct torque, proper base/grout. A misaligned coupling can cut bearing life by an order of magnitude before the machine has run a week.
• Lubrication discipline – right lubricant, right quantity, right interval, kept clean and dry. Lubrication issues are consistently among the top contributors to premature bearing failure (see our SKF summary).
• Operating within design envelope – cavitation, dead-heading, overloads and frequent starts all consume the I-P interval silently.
• Defect elimination culture – treating every early detection as a question ("what put this defect in?") instead of a victory lap for the vibration analyst.