Skip to content
allanninal.dev/build
Collaborate

Part 2 of 7 · Equipment maintenance scheduler series ~5 min read

How the interval gets decided

Most maintenance is scheduled monthly because months are easy to put in a spreadsheet, and for equipment whose use varies that produces both unnecessary servicing and missed servicing at the same time.

Key takeaways

  • Running hours or cycles beat calendar months wherever the machine counts them.
  • Whichever comes first: some things degrade with time even when idle.
  • Machines that count nothing get a proxy — output, shifts run, or a manual reading.
  • Servicing too often costs money and introduces failures of its own.
  • Say on every job which basis it was scheduled on.

Usage against calendar

Running hours over six months for three identical machinesA bar chart with three bars showing running hours over six months. Machine A, busy: six hundred and twenty hours. Machine B, average: four hundred and ten hours. Machine C, a spare: ninety-five hours. A note says on a six-month schedule all three get one service, while A needs two and C needs none.0200400600800~620Machine A, busy~410Machine B, average~95Machine C, spareRunning hours in six monthsOn a six-month schedule all three get one service. A needs two; C needs none.
Fig 1. Three identical machines over the same six months. A calendar schedule treats them the same and is wrong about two of them in opposite directions.

Machine C is the interesting case. Servicing it on the calendar means paying for work that is not needed, and every intervention carries a small risk of introducing a problem that was not there. Servicing lightly-used equipment on schedule is not caution; it is cost with a failure mode attached.

Whichever comes first

Usage alone is not sufficient either, because some things degrade with time regardless: rubber perishes, oil oxidises, seals dry out, batteries self-discharge. The practical rule is an interval in hours and a maximum in months, whichever arrives first.

Machine C then gets serviced annually rather than never, on the calendar limit, and the job record says which limit triggered it. That distinction is worth keeping because a machine that is always triggered on calendar is a machine you may not need.

Machines that count nothing

How the scheduling basis is chosen for a piece of equipmentA vertical chain of five steps entered by a box labelled A machine to schedule, asking what it reports. Step one asks whether there is an hour meter, which most plant has; if so it exits to Read it, automatically or weekly. Step two asks whether there is a cycle counter for presses, doors or pumps; if so it exits to Use cycles, often better than hours. Step three asks whether a proxy is available such as output, shifts or distance; if so it exits to Use it and say so, approximate rather than fake. Step four falls back to calendar only, with a stated assumption. Step five records the basis on every scheduled job. A note says a proxy that is honest about being a proxy beats a precise number that is invented.AWS ACCOUNTA machine to schedulewhat does it report?An hour meter?most plant has oneRead itautomatically or weeklyyesA cycle counter?presses, doors, pumpsUse cyclesoften better than hoursyesA proxy available?output, shifts, distanceUse it, and say soapproximate, not fakeyesCalendar onlywith a stated assumptionRecord the basison every scheduled jobA proxy that is honest about being a proxy beats a precise number that is invented.
Fig 2. How the scheduling basis is chosen per machine. The last box is what stops a proxy quietly becoming treated as a measurement.
  • Compute
  • App integration
  • Management
  • Analytics

Cycles beat hours sometimes

For anything that starts and stops — a press, a compressor, a door, a pump — the wear is in the cycles rather than the hours. A compressor running continuously for eight hours does less damage to itself than one starting forty times in the same period.

Where both are available, the interval should be on whichever the manufacturer specifies and the other recorded alongside, because the ratio between them is itself diagnostic. A compressor whose cycles per hour have doubled has a leak somewhere.

Over-maintenance is real

The costs of servicing too often

  • The direct cost: parts and labour for work that was not needed.
  • The downtime: a machine stopped for a service it did not need is a machine stopped.
  • Intervention failures: a proportion of failures happen shortly after maintenance, because somebody was inside the machine.
  • The crying wolf effect: a schedule full of unnecessary jobs teaches people that the schedule is negotiable.
  • The right response is to lengthen intervals where the evidence supports it, recorded as a deliberate change with a reason.
  • Not to quietly skip jobs, which produces the same saving and none of the knowledge.

The third line is the one people find counter-intuitive and it is well established. Opening a machine up, disturbing settled assemblies and reassembling them introduces a small failure rate, which is why the interval should be as long as the evidence supports rather than as short as caution suggests.

This is also the argument for recording completed jobs with what was found. A service where nothing needed replacing, three times running, is evidence that the interval could be longer, and that evidence only exists if somebody wrote down that nothing was wrong.

Next: what happens when a job gets put off.

All posts