8.3 Preventive and Predictive Maintenance

Why This Topic Matters

Module 15508 is the capstone module of the Maintenance and Troubleshooting domain: it ties together everything covered elsewhere in this guide — bearings, lubrication, alignment, pumps, gearboxes — into a maintenance strategy. Rather than testing one machine type, this topic tests whether a millwright understands which condition-monitoring technique detects which kind of developing fault, which is exactly the "match the tool to the symptom" question style the exam favors here. Expect matching-style items: given a described problem (a loose electrical connection, a developing bearing defect, contaminated oil), identify the technique that would catch it earliest.

Maintenance Strategies Compared

Reactive (breakdown) maintenance repairs equipment only after it fails. It has the lowest planned cost per event but the highest total cost, since failures happen at the worst time, cause secondary damage, and create unplanned downtime. Preventive maintenance (PM) performs scheduled tasks — lubrication, filter changes, belt inspections, bearing replacement at a fixed interval — regardless of the equipment's actual condition, based on calendar time or usage (hours, cycles). PM reduces surprise failures but can waste good parts (replacing something that didn't need it yet) or still miss a fault that develops faster than the schedule assumes. Predictive maintenance (PdM) is condition-based: it uses measured data — vibration, oil condition, temperature, sound — trended over time, so intervention happens only when the equipment's actual condition calls for it, ideally before failure but without discarding remaining useful life.

The P-F Curve

As a component degrades, there is a window between the point a fault becomes detectable by some monitoring technique (the "potential failure," or P point) and the point it causes an actual functional failure (the F point) — this interval is called the P-F curve. The entire purpose of predictive maintenance technology is to detect faults as early as possible on this curve, maximizing the lead time available to plan a repair during a scheduled outage instead of reacting to a breakdown. Different PdM technologies detect faults at different points on the curve — for example, ultrasonic testing can often pick up early bearing lubrication starvation before it produces enough vibration energy for a vibration analyzer to flag it.

Predictive Maintenance Technologies

Vibration analysis is the primary PdM tool for rotating equipment. An accelerometer captures overall vibration and breaks it into a frequency spectrum: vibration at 1X running speed typically points to imbalance, 2X running speed commonly points to misalignment or mechanical looseness, and bearing-defect frequencies (tied to the bearing's specific geometry) show up at frequencies that are not simple multiples of running speed. Overall vibration severity is judged against reference standards such as ISO 10816/20816, which define zones from newly-commissioned/acceptable through requiring-action to dangerous, with the specific mm/s thresholds depending on the machine's size and mounting class.

Lubricant (oil) analysis checks a sample against baseline for viscosity (a shift outside spec suggests the wrong oil, oxidation, or contamination), Total Acid Number (TAN) (a rising TAN indicates oxidation of the base oil), water contamination, and wear-metal spectroscopy — iron in the sample points to gear or bearing wear, copper points to bushing/bronze component wear, and silicon points to dirt/abrasive ingress.

Infrared thermography uses a thermal camera to find components running hotter than a comparable component or their own historical baseline under similar load — commonly used to find loose electrical connections at motor control centers and switchgear, overheating bearings, and insulation breakdown, since a meaningful comparison always has to be load-normalized rather than a single absolute number.

Ultrasonic testing detects high-frequency sound inaudible to the human ear, making it effective for finding compressed-air and gas leaks, electrical arcing/corona/tracking on energized equipment, early-stage bearing friction from lubrication starvation, and steam trap failures.

Nondestructive testing (NDT) methods that a millwright should recognize even when a specialist performs them include visual inspection (VT), dye penetrant testing (PT) for surface cracks, magnetic particle testing (MT) for surface/near-surface cracks in ferrous material, ultrasonic testing (UT) for internal flaws and thickness measurement, and radiographic testing (RT) for internal defects using X-ray or gamma sources.

Building a PM/PdM Program

A working program relies on a CMMS (Computerized Maintenance Management System) to schedule work orders, track equipment history, and store inspection/condition data; PM schedules driven by calendar interval or usage meter; pre-commissioning and startup checks on new or rebuilt equipment (alignment verification, rotation check, and — critically — recording an initial vibration and thermal baseline while the equipment is known-healthy, since every later trend comparison depends on having that reference point); and root cause analysis (RCA) after any failure, so the same failure mode doesn't simply repeat on the next PM cycle. Larger, more critical assets are also managed under a reliability-centered maintenance (RCM) framework, which formally weighs an asset's criticality and failure modes to decide whether reactive, preventive, or predictive maintenance (or some mix) is the right strategy for that specific piece of equipment.