Types of Maintenance

The options

• Reactive (corrective) — repair after failure. Appropriate where failure is inconsequential, cheap to fix, and causes no secondary damage. Inappropriate, and expensive, everywhere else.
• Preventive (planned) — scheduled work at fixed intervals of time or usage, regardless of condition. Simple to plan and resource, and it replaces components that still had life.
• Predictive (condition-based) — monitoring condition and intervening when indicators show deterioration. Maximises component life and avoids failure, at the cost of monitoring equipment and the skill to interpret it.
• Statutory — inspections and tests required by law, at defined frequencies, by competent persons. Non-negotiable regardless of condition or cost.
• Improvement — modifying the asset or its environment to remove the failure mode entirely, which is the only intervention that permanently reduces work.

The true cost of reactive maintenance

Reactive maintenance appears cheap because the only visible cost is the repair. The actual cost includes: the consequential loss while the asset is unavailable; secondary damage caused by the failure; emergency call-out and premium parts pricing; disruption to occupants; overtime and disrupted planning for the maintenance team; and shortened asset life from running to failure. Organisations that measure only labour and parts systematically underestimate what reacting costs them.

Choosing per asset

No single strategy fits everything. Decide asset by asset, considering:

• Criticality — what happens if it fails, how quickly and to whom.
• Failure pattern — whether failure is age-related and therefore predictable by interval, or random, in which case interval-based replacement achieves nothing.
• Detectability — whether deterioration gives warning that can be monitored.
• Cost — of the maintenance, against the cost of the failure.
• Redundancy — whether there is standby capacity.
• Statutory requirements, which override everything else.

The failure pattern point

A widely cited finding in reliability engineering is that only a minority of failure modes are age-related; most exhibit random or early-life failure patterns. Where failure is random, replacing a component at a fixed interval does not reduce failures and may increase them, because the intervention itself introduces early-life failure risk. This is the strongest argument for condition monitoring over interval replacement on complex equipment, and for not opening up equipment that is working correctly.

Reliability-centred thinking

A structured approach asks, for each asset: what are its functions, how can each function fail, what causes each failure, what happens when it does, how much does that matter, and what task would prevent or detect it? If no worthwhile preventive task exists, the honest answers are to run to failure deliberately, to add redundancy, or to redesign. Applying this formally to every asset is expensive; applying the logic to the critical ones is valuable and achievable.