Maintenance management: running the function that keeps the plant available

Choosing a different strategy for different assets

No plant can afford intensive maintenance on everything, and no plant survives running everything to failure. The allocation should follow consequence: assets whose failure stops the plant, endangers people or ruins product justify planned intervention and monitoring, while assets with a spare, a quick swap or no production consequence can reasonably be run until they fail. Making that decision explicitly, asset by asset, is what separates a managed function from a reactive one. Record the chosen strategy against each asset and revisit it when the production role of the equipment changes, because yesterday's spare machine becomes today's constraint.

The work order flow from request to closure

Every intervention should start as a request that someone screens: is it real, is it a duplicate, what is its priority, does it need a shutdown, what parts and skills does it need. Unscreened requests go straight to a backlog nobody trusts. After screening comes preparation, scheduling with production, execution and a closing record of what was found and done. The closing record is the part most often skipped and the most valuable, because a history of what actually failed is the only basis for deciding whether the current strategy for that asset is right.

Backlog is a management instrument, not a failure

A maintenance backlog of zero means the function is overstaffed or that requests are being discarded; an unbounded backlog means requests disappear into it and people stop raising them. What matters is that the backlog is sized in work hours rather than in job count, sorted by consequence, and reviewed at a routine where production and maintenance jointly agree what is deferred. Ageing matters too: a safety-related or compliance job sitting untouched is a different problem from a cosmetic one. Report the shape of the backlog, not just its total, and make deferral a recorded decision with a name against it.

Defending planned work against the pressure of the day

The permanent tension is that planned maintenance requires equipment time that production always wants. When the plant is behind, the planned job gets cancelled, which raises the failure rate, which puts the plant further behind. Breaking that loop needs a rule agreed above both functions: which planned tasks may be deferred, by whom, for how long, and what happens when deferral repeats. Tracking the proportion of planned work completed as scheduled makes the erosion visible early. Where the figure collapses whenever demand rises, the plant is borrowing availability from its future and should say so openly in its capacity planning.

Measuring the function on outcomes rather than activity

Counting completed jobs or hours worked rewards busyness. Better measures look at the outcome for production: unplanned downtime on the assets that matter, the proportion of work that was planned rather than reactive, repeat failures on the same asset, and the time from a stoppage being raised to the equipment running again. Maintenance cost should be reported alongside these rather than alone, because a cost reduction achieved by deferring work looks like an improvement for a while. Present the set together to the plant leadership, since the trade-off between spend and availability is a business decision, not an engineering one.

Frequently asked questions

Should maintenance report to production or to engineering?

Both structures work and both have a characteristic failure. Under production, planned work gets sacrificed to output pressure and the asset base degrades slowly. Under engineering, maintenance can become detached from what the plant actually needs today and slow to respond. Whichever is chosen, the safeguards are the same: a jointly agreed rule on deferring planned work, a shared review of the backlog, and availability reported to a level above both functions.

How much should a plant spend on maintenance?

There is no defensible universal ratio, because it depends on asset age, duty, criticality and the cost of an hour of lost output. A more useful approach is to work from consequence: establish what an hour of downtime costs on the assets that limit the plant, then judge whether current spending is buying enough availability on those specific assets. Benchmarks expressed as a share of asset value tend to justify whatever the plant is already doing.

What does maintenance need from production to work properly?

Access to equipment at agreed times, honest reporting of symptoms before failure rather than after, a stable enough production plan that shutdown windows can be scheduled in advance, and involvement in decisions about running equipment outside its intended duty. In return production is entitled to notice of planned work, realistic duration estimates and clear communication when a job overruns. Most disputes between the two functions trace to one of these being absent rather than to technical disagreement.