Preventive Maintenance: A Risk-Based Industrial Guide

What preventive maintenance should achieve

The purpose of PM is to control known failure mechanisms before they create unacceptable consequences. Examples include loss of lubrication, filter blockage, belt wear, loose electrical connections, seal degradation, fouling, corrosion and loss of a protective function. Every task should answer two questions: which failure mode does this address, and what evidence shows the task is effective?

Preventive maintenance is broader than calendar replacement. It may include inspection, cleaning, lubrication, adjustment, functional testing and planned replacement. Condition-based maintenance uses an observed condition to decide when action is required. Predictive maintenance uses condition history or a model to estimate future deterioration. These approaches belong in the same strategy; the correct choice depends on failure behavior, consequence, detectability and available evidence.

A seven-step risk-based planning workflow

1. Define the asset and required function. Record location, duty, operating context, manufacturer, model and the performance the asset must deliver.
2. Rank consequence and criticality. Consider safety, environment, production, quality, service continuity, compliance and repair exposure. A high consequence justifies more rigorous analysis, not automatically more calendar work.
3. Identify credible failure modes. Use OEM guidance, breakdown history, operator observations, technician knowledge, inspection findings and a structured maintenance FMEA where useful.
4. Select the maintenance tactic. Match the task to how the failure develops. Do not assign a routine task merely because the asset is important.
5. Write the task and acceptance criteria. State the maintenance condition, method, unit, expected range or pass/fail evidence, escalation route and close-out record. The professional PPM checklist guide shows how to turn this analysis into technician-ready work.
6. Choose an initial interval. Start with approved OEM, design, statutory and site requirements, then consider duty, environment, redundancy, usage and known deterioration.
7. Review results. Use findings, failures, trends and work quality to keep, change or retire the task through a controlled approval process.

Choose a tactic that matches the failure

• Time- or usage-based replacement: useful when age, starts, cycles or operating hours have a meaningful relationship with failure and a defined replacement policy exists.
• Condition inspection: useful when deterioration is detectable early enough to plan a safe intervention. Examples include leakage, contamination, wear, temperature or vibration trends.
• Functional test: useful for standby or protective functions whose failure may remain hidden during normal operation.
• Restoration or servicing: useful when cleaning, lubrication, adjustment or calibration demonstrably restores required condition or performance.
• Run to failure: may be rational only when consequence is acceptable, failure is evident, recovery is manageable and no cost-effective preventive task exists.
• Redesign or operating change: required when routine maintenance cannot control the risk or the same failure keeps returning.

Use condition monitoring as a decision process

A measurement is not useful simply because it can be collected. Define what is measured, where and how it is measured, the unit, operating state, baseline, trend direction and action route. Comparable readings matter: a temperature, current or vibration value taken at a different load or location may not support a valid conclusion.

Use approved site or OEM limits where they exist. When they do not, establish a controlled baseline from healthy operation and review trends with competent technical judgment. One abnormal reading may require verification; a worsening trend may justify planned investigation. Never invent a universal alarm threshold for equipment with different designs, duties or measurement methods.

Condition monitoring does not replace basic work quality. Sensors and routes cannot compensate for poor identification, inconsistent measurement points, missed defect escalation or unsafe access. Record findings in a form that can generate corrective work, not as unstructured “checked OK” comments.

Set and change frequencies with evidence

An initial frequency is a hypothesis about how quickly deterioration can become unacceptable. Mandatory OEM, statutory and approved site requirements take priority. Within those boundaries, review the interval against duty, operating environment, failure-development time, detectability, inspection quality and opportunity to intervene.

• If defects repeatedly appear between tasks, check the failure mode, task quality and tactic before simply shortening the interval.
• If many cycles find no deterioration, review whether the task is low value, the interval is conservative or another tactic is better.
• If a finding creates no corrective action, confirm that the acceptance criteria and escalation ownership are clear.
• If operating duty, process conditions or equipment design change, reassess the task rather than carrying the old frequency forward.

Frequency changes should be documented with the evidence, risk review, approver and next review date. Avoid optimizing a schedule from a small sample or a period with unusually light duty.

Worked example: motor-driven process pump

Assume a duty pump has repeated coupling wear. “Inspect pump monthly” is too vague. The team first defines credible causes: misalignment, pipe strain, soft foot, looseness, unsuitable coupling selection or poor installation. It then selects tasks that can detect or control those causes: observe vibration and leakage under a defined operating condition, check foundation and fastener condition, confirm alignment after intrusive work, and record measured evidence against approved criteria.

If the same defect returns, increasing the inspection frequency may only detect the symptom more often. The better response may be a root-cause investigation, installation correction or redesign. This example shows the boundary between preventive work and problem elimination.

Plan the maintenance state and safe execution

Define whether each task is performed running, stopped, isolated, depressurized, cooled or dismantled. Observation during operation and intrusive maintenance after isolation are different tasks with different hazards and evidence. The work instruction should specify access, tools, consumables, competence, restoration checks and the required record.

Measure reliability, not checklist volume

PM completion percentage is a delivery measure, not proof that the strategy works. Review it alongside repeat failures, emergency work, schedule compliance, PM-generated corrective actions, backlog age, defect recurrence, availability, MTBF, MTTR and failure cost. Keep boundaries consistent before comparing periods or assets.

Separate three questions during review: was the correct task scheduled, was it executed to the required quality, and did the chosen tactic control the failure? This prevents a weak strategy from being hidden by a high completion percentage.

Beginner → Intermediate → Advanced progression

1. Beginner: map assets, functions, failure modes and basic task evidence. Then use the PPM checklist guide and blank CSV template to write clear technician tasks.
2. Intermediate: rank criticality, apply FMEA, standardize condition routes and connect findings to corrective work.
3. Advanced: study the curated reliability engineering video pathway, optimize intervals with sufficient history, compare failure and exposure data using the MTBF/MTTR worksheet, and test decisions with the reliability calculator.

A practical review loop

1. Review major failures, repeat defects, overdue critical work and PM-generated corrective actions.
2. Confirm whether the failure mode and consequence were understood.
3. Check task clarity, execution evidence and maintenance state.
4. Decide whether to retain, rewrite, reschedule, replace or retire the task.
5. Assign an owner, approval and follow-up date, then verify the outcome.

Reliability improves when PM is treated as an engineering feedback loop rather than a fixed checklist. The goal is less avoidable risk and better asset performance—not the largest possible task library.

Related engineering guides