15.3 Three Maintenance Modes

Classify work by its trigger

The WPI outline explicitly requires corrective, preventive, and predictive maintenance. The U.S. Environmental Protection Agency's 2024 Effective Utility Management primer describes preventive and predictive work as planned maintenance: preventive work follows a predetermined schedule, while predictive work begins when signals show that maintenance is due. Other work is corrective or reactive. For the exam, use that trigger-based distinction rather than assuming that every planned job is preventive.

Preventive maintenance (PM) occurs at a calendar, runtime, cycle, or other prescribed interval. Examples include an inspection scheduled by operating hours, replacement required at a manufacturer interval, or an exercise on the utility's approved calendar. PM aims to reduce known failure mechanisms, but more maintenance is not automatically better. An unnecessary intrusive task can introduce contamination, misalignment, wrong lubricant, or assembly error. The interval and scope come from manufacturer guidance, operating experience, regulation, and the site's approved program.

Predictive maintenance (PdM) uses condition evidence to decide when intervention is warranted. The signal might be a vibration trend, thermographic finding collected by qualified staff, motor-current pattern, lubricant analysis, changing differential pressure, declining capacity at comparable load, or repeated temperature deviation. A single reading without context is weak. The method, sensor location, units, load, speed, and environment should be comparable to the baseline. Predictive does not mean a computer predicts an exact failure date; it means condition informs the timing and scope of work.

Corrective maintenance (CM) repairs or adjusts a known defect and restores required function. It can be an emergency response after failure, a nonemergency repair of degraded equipment, or planned correction created from a predictive finding. For example, a condition route can detect worsening bearing behavior, which generates a scheduled corrective work order before the blower fails. The diagnostic activity is predictive; the repair is corrective.

Turn observations into defensible decisions

A practical condition route asks five questions: What asset and component were observed? Under what operating state? With what method and units? How does the result compare with baseline and prior values? What process consequence follows if the trend continues? Record normal findings as well as exceptions; otherwise the history cannot show when change began. Never invent a universal vibration, temperature, insulation, oil, or current limit. Use the approved alert and action criteria for that machine and method.

When an abnormality appears, first validate the observation. Confirm the asset tag, instrument status, measurement point, units, and load. Look for a process explanation, such as a valve change that altered blower duty. Compare redundant equipment where meaningful. Then assess risk: safety or environmental consequence, effect on treatment, likelihood and pace of failure, available redundancy, ability to monitor, spare and contractor lead time, and required outage. Immediate hazards follow the emergency and safe-shutdown procedure. A stable low-consequence defect may be planned, but it must not disappear into an informal notebook.

The condition-to-work chain

1. Detect: an operator round, alarm, test, or analysis identifies a change.
2. Validate: confirm the reading and operating context without bypassing protection.
3. Assess: combine severity and trend with criticality and redundancy.
4. Plan: define scope, hazards, isolation, permits, people, parts, process contingency, and acceptance criteria.
5. Execute: authorized personnel perform the controlled work and record actual findings.
6. Verify: return the asset under the approved procedure and confirm both equipment and process performance.
7. Learn: update history, task frequency, failure coding, spares, and operating guidance when the evidence supports a change.

Planning is a reliability control. A strong work package says which asset and symptom, what function must be restored, what energy and process hazards exist, what parts and skills are required, how treatment continues during the outage, and how success will be tested. “Check blower” is not an adequate scope. The person scheduling work also considers whether two redundant units would be made unavailable at once.

Scenario: rising vibration, no alarm

Three comparable readings show a blower's vibration trend increasing while air delivery at the same operating condition slowly declines. The displayed value has not crossed the plant alarm. Treating the machine as healthy merely because no alarm exists wastes predictive evidence. The operator verifies the measurement context, checks for an accessible process restriction or changed valve state, records the trend, and requests the facility's authorized assessment. Criticality, rate of change, standby readiness, and the next available outage determine priority.

Suppose specialists find coupling misalignment and schedule correction. The vibration monitoring remains predictive; correcting the confirmed defect is planned corrective maintenance. If the plant later updates an alignment check or inspection interval based on recurring evidence, that future scheduled task is preventive. One event can therefore touch all three modes without making the terms interchangeable.

After work, return-to-service verification closes the loop. Confirm guards and work clearance under the SOP, then observe startup through normal interfaces and compare condition and process output with acceptance criteria. Record the result, including an unsuccessful result. A closed work order that says only “done” prevents learning and may conceal repeat failure. The exam-ready principle is simple: classify the trigger correctly, use trends rather than guesses, plan around treatment risk, and verify restoration.