Where maintenance intervals should come from

Ask why a task runs quarterly and the answers cluster into three: it is a legal requirement, the manufacturer says so, or it has always been quarterly. Only the first is genuinely fixed. For a separate product-side perspective on the limits of self-reported operational data, this glossary entry provides additional detail.

Distinguishing between them is the highest-leverage exercise available in maintenance planning, because it identifies which intervals are negotiable and which are not.

Statutory intervals are fixed

Certain inspections are required by law or by insurance, at frequencies set externally: pressure systems, lifting equipment, electrical installations, fire safety systems, gas appliances, water systems. The specific requirements and frequencies differ substantially between jurisdictions. For a wider operational and compliance reference, consult the Better Buildings O&M guide.

These are not candidates for optimisation. They are a floor, they usually require a competent or accredited person, and the record is as important as the inspection — an inspection that happened but was not documented has, for most practical purposes, not happened.

Manufacturer intervals are a starting point

Manufacturer recommendations are set for the manufacturer's risk, not yours. They assume a duty cycle, an environment and a consequence of failure, generally at the demanding end of all three, and they are written with warranty exposure in mind.

For an asset running at a fraction of rated duty in a clean plant room, the recommended interval may be substantially conservative. For one running continuously in a dusty or corrosive environment, it may be optimistic. Neither is a reason to ignore it; both are reasons to treat it as an input rather than an answer — noting that departing from it may affect warranty during the warranty period.

Let criticality do the sorting

The consequence of failure should drive the frequency more than the asset type does. Two identical pumps deserve different treatment if one has a standby and the other does not.

The practical version is a small number of criticality bands, each with a default posture: high criticality gets shorter intervals and often condition monitoring; medium gets the standard interval; low criticality is a legitimate candidate for run-to-failure, provided that is an explicit decision with a spares position behind it rather than neglect that has been rationalised.

Use your own failure history

After a few years, the maintenance record contains the most relevant evidence available: what actually failed, when, and whether a planned visit had recently occurred.

Two patterns are worth looking for. Tasks that repeatedly find nothing across many cycles are over-frequent. Assets that fail shortly after a planned visit point either to an interval that is too long or to a task that is not inspecting the thing that fails — and the second is more common than the first.

Condition monitoring changes the question

Where a meaningful condition can be measured — vibration, temperature, current draw, run hours, differential pressure across a filter — the interval question partly dissolves. The task runs when the measurement says so rather than when the calendar does.

This is genuinely better for suitable assets and is not free: it requires the measurement to be taken reliably, a threshold that has been thought about, and somebody who will act when it is crossed. Condition data that nobody reviews is worse than a calendar interval, because it creates the impression of monitoring.

Write down why

Whatever interval is chosen, record the reason in one line against the task: statutory, manufacturer, criticality-based, or history-based, with the date of the decision.

This is what makes the schedule reviewable. Without it, every interval looks equally fixed to whoever inherits the system, and the whole set becomes untouchable — which is exactly how the current unexamined schedule came into existence.