Complete sample module, free to use

Failure Modes & the P-F Curve

Course
PM-01
Module
01
Contact hours
3
Version
1.0

This is a complete sample module, given away in full. Use it, teach it, adapt it. It is the first unit of a predictive maintenance program, and it exists here so you can judge the quality of the work before any conversation about the rest.

In this module
  1. Why this module comes first
  2. 1. The four maintenance strategies
  3. 2. The P-F curve
  4. 3. Failure modes and early signs
  5. 4. Lab: Diagnose the interval
  6. 5. Assessment
  7. 6. Instructor notes

Why this module comes first

Most people arrive at predictive maintenance believing it means better sensors. It doesn't. Predictive maintenance is a bet on time. You detect a developing fault early enough to act before the equipment fails, and everything downstream follows from understanding how a particular machine actually fails.

Get that wrong and the rest of the program becomes button-pushing: readings collected on a route, alarms nobody trusts, and a dashboard open on a monitor that no one looks at. This module is the conceptual keystone. Teach it properly and the technique modules land. Skip it and they don't.

On completion, a participant can

  • Distinguish reactive, preventive, predictive and proactive maintenance, and say when each is the economically correct choice
  • Read and construct a P-F curve, and explain why the P-F interval determines monitoring frequency
  • Identify the dominant failure modes in rotating equipment and their early symptoms
  • Map a given failure mode to the condition-monitoring technique most likely to catch it early

The four maintenance strategies

Every maintenance decision is a choice among four postures. None is universally right; the skill is knowing which the asset deserves.

Table 1. The four maintenance strategies
StrategyWhat it meansWhen it's correct
ReactiveRun the asset until it fails, then repair.Cheap, non-critical, redundant assets where failure costs less than monitoring.
PreventiveService on a fixed calendar or runtime schedule.Known wear-out patterns, regulatory requirements, low-cost interventions.
PredictiveMonitor condition, act when evidence of a developing fault appears.Critical assets with a detectable, gradual failure path.
ProactiveEliminate the root cause so the failure mode stops recurring.Repeat failures where the underlying cause is fixable: alignment, lubrication, design.

Teaching point

Participants often assume predictive is always superior. It isn't. Monitoring a $200 pump with a spare on the shelf is a waste of a route stop. Criticality drives the choice.

The P-F curve

Equipment rarely fails instantly. Condition degrades over a period, and somewhere along that decline the fault becomes detectable. That is point P. Later, the asset can no longer do its job. That is point F, functional failure. The time between them is the P-F interval, and it is the single most useful number in the discipline.

The P-F curve Equipment condition stays level, then declines over time in service. Point P, marked with a circle, is where the fault becomes detectable. Point F, marked with a square, is functional failure. The hatched span between them is the P-F interval. Condition Time in service PFault becomes detectable FFunctional failure P-F intervaltime to detect, decide, and act The P-F curve Equipment condition stays level, then declines over time in service. Point P, marked with a circle, is where the fault becomes detectable. Point F, marked with a square, is functional failure. The hatched span between them is the P-F interval. Pdetectable Ffailure P-F interval time to detect, decide, act
The interval is not fixed. It depends on the failure mode and on which technique you use to look for it. A more sensitive technique moves P to the left and buys more time.

The rule that follows

Monitoring frequency

To catch a fault reliably, you must inspect more than once within the P-F interval. The common working rule is to monitor at roughly half the interval or less.

A bearing fault with a 30-day detectable window inspected quarterly will be found by the breakdown, not by the route.

This is where most programs quietly fail. The technique is fine, the sensor is fine, and the inspection frequency was set by tradition rather than by the interval.

Common failure modes and their early signs

Table 2. Failure modes, earliest signals, and the technique that sees each first
Failure modeEarliest reliable signalTechnique that sees it first
Bearing degradationHigh-frequency energy well before audible noiseVibration; ultrasonic
MisalignmentCharacteristic axial vibration patternVibration
ImbalanceElevated vibration at running speedVibration
Lubrication failureWear particles and contamination in the oilOil analysis
Electrical connection faultLocalised heat rise under loadThermography
Compressed air / steam leakUltrasonic emission at the leak pointUltrasonic
Motor winding / rotor faultSidebands in the current signatureMotor current analysis

Teaching point

The right-hand column previews the rest of the program. Each technique exists because it sees something the others miss. That is the argument for a mixed program rather than buying one tool.

Lab exercise: "Diagnose the interval"

Participants work in pairs. For each scenario: name the likely failure mode, estimate where it sits on the P-F curve, choose a monitoring technique, and set an inspection frequency. Allow 35 minutes, then debrief as a group.

  • Pairs
  • 35 minutes
  • Group debrief
  1. 1Process pump

    A centrifugal pump on a critical line has been rebuilt twice in eighteen months, each time for bearing failure. Nothing else has changed.

  2. 2Motor and gearbox

    A gearbox coupled to a 75 hp motor runs slightly warmer than its neighbours. Output is unchanged and there is no unusual noise.

  3. 3Conveyor drive

    A packaging line conveyor has begun tripping its overload intermittently, roughly once a week, always in the afternoon.

  4. 4Air compressor

    The plant's compressed air system now runs a duty cycle nearly 20% higher than a year ago. Production volume is flat.

  5. 5Exhaust fan

    A rooftop exhaust fan, non-redundant but not process-critical, has a replacement cost of about $1,800 installed and a two-day lead time.

  6. 6Switchgear

    A distribution panel serving three production cells has never been inspected beyond visual checks. It is fifteen years old.

Debrief: the point of the exercise

Several scenarios have more than one defensible answer, and that is deliberate. Scenario 5 is the trap: the correct answer is usually no monitoring at all. A cheap, quickly replaced, non-critical asset does not earn a route stop. Participants who monitor everything have missed the economics. Scenario 4 is a system-level symptom rather than a single failing machine, and rewards anyone who reasons about the whole air system rather than one compressor. The exercise teaches judgment, not lookup.

Assessment

Ten multiple-choice items plus one short-answer question. The short-answer item is the one that matters:

Short answer

A bearing on a critical pump has a P-F interval of approximately 30 days. Your current route inspects this asset quarterly. Explain what is wrong with the present arrangement and what you would change, including your reasoning about frequency.

Expected response

Quarterly inspection is roughly a 90-day cycle against a 30-day detectable window, so the fault will almost always progress to failure between inspections. The monitoring provides no useful warning and the cost of the route is wasted. Frequency must fall inside the interval, conventionally at half or less, so a fortnightly or monthly check is defensible. A stronger answer also raises continuous monitoring if the asset's criticality justifies the sensor cost.

Instructor notes

Timing
Roughly 40 minutes lecture, 35 minutes lab, 25 minutes debrief, 20 minutes assessment and review, with the balance for questions.
Common misconceptions to correct
That predictive maintenance is always the best strategy; that the P-F interval is a fixed property of a machine rather than of the failure mode and technique used; and that more sensors means more warning, when sensors nobody responds to produce no warning at all.
The one thing to land
If participants leave with only one idea, it should be that inspection frequency must be set by the P-F interval, not by tradition or convenience. Nearly every disappointing predictive maintenance program in the field can be traced back to that single error.
Adapting the length
For a two-hour block, drop scenarios 3 and 6 and shorten the debrief. For a full-day session, add a plant walk in which participants rank ten real assets by criticality before the lab.

Use this freely.

This module is yours to teach, copy and adapt. No registration, no follow-up sequence, no salesperson. If it's useful, the full PM-01 course runs six modules and can be licensed for your program, or taught.