Machine utilisation: what the figure means and how it misleads people

Say which denominator you are using, out loud

Against total calendar hours, most equipment looks idle and any single-shift plant appears to have vast latent capacity. Against scheduled hours, the figure reflects how well the plant uses the time it chose to operate. Against manned hours, it reflects how well an operator's presence is converted into running time. Each answers a different question — the first informs whether to add a shift, the last informs whether to add a machine — and quoting one while the audience assumes another produces confident, expensive mistakes. Write the definition next to the number every time it is published.

Utilisation, availability and the conflation that follows

Availability describes the proportion of intended running time the machine was capable of running: it is limited by breakdowns, maintenance and setup. Utilisation describes how much of that time was actually spent producing, which is additionally limited by whether there was work, material and an operator. A machine can be entirely available and barely utilised because the order book is thin, and that distinction determines who owns the problem — engineering for availability, planning and sales for utilisation. Combined efficiency measures that fold in rate and quality belong to improvement practice; here the split matters for accountability.

Away from the constraint, high utilisation is a cost

Running a non-constrained machine hard produces inventory the next operation cannot consume, consumes material and labour earlier than needed, and hides the real limitation. It also feels like good management, which is why it persists. The only place utilisation should be pushed toward its ceiling is at the resource that actually limits plant output, and even there the goal is uninterrupted flow rather than a percentage. Anywhere else, idle time on a machine is not waste — it is the flexibility that lets the plant respond when the mix changes, and paying for it is usually deliberate.

What the figure is legitimately good for

Three uses stand up. Capacity decisions: whether the plant needs another machine, another shift or a subcontractor, which needs utilisation measured against realistically available time. Quoting and costing: hourly rates depend on assumed utilisation, and an optimistic assumption underprices work systematically. Investment review: whether a machine bought two years ago is doing what the case claimed. Outside those, publishing utilisation by machine as a performance league table produces exactly the behaviour described above, and the plant pays for it in inventory. Subcontract decisions are a fourth borderline case: knowing how loaded a work centre genuinely is tells you whether sending overflow outside is relieving a real constraint or simply paying somebody else to hold your idle time.

The behaviour a utilisation target reliably produces

Set a target and people will meet it. Operators run long batches to avoid changeovers, supervisors pull work forward to keep a machine loaded, planners release orders early so nothing stands idle, and maintenance windows get deferred because downtime hurts the figure. Every one of those is rational given the target and damaging to the plant. If utilisation is to be measured at all outside the constraint, publish it alongside work in progress and delivery performance, so the cost of pushing one is visible in the others. Otherwise the measure will be optimised at the expense of the thing it was meant to serve.

Frequently asked questions

What utilisation level should a well-run plant aim for?

There is no universal answer, and pursuing a single figure across a mixed plant guarantees the wrong behaviour. Constrained equipment should be loaded as fully as the schedule and maintenance allow. Equipment whose job is to absorb variation should deliberately carry slack, because that slack is what protects delivery when the mix shifts. Judge each work centre by whether it constrains output, then set expectations accordingly rather than applying one number to everything.

Should setup time count as utilised or idle?

Count it separately, as its own category, and stop the argument. Folding setup into utilised time flatters the machine and hides the changeover burden; counting it as idle makes a high-variety cell look catastrophic. Reporting running, setup, planned stoppage and unplanned stoppage as four distinct buckets summing to the scheduled time gives a picture people can act on, and it lets setup reduction be judged on its own evidence.

Can utilisation figures justify buying another machine?

Only with the denominator stated and the constraint identified. High utilisation on a machine that is not limiting output justifies nothing. High utilisation at the constraint, sustained across the realistic operating pattern with maintenance and setup accounted for, is a genuine signal — though the alternatives should be tested first: another shift, a changeover reduction, subcontracting the overflow, or moving work to underused equipment. Capital is usually the most expensive of those options and the easiest to approve.