← The journalMANUFACTURING / FIELD GUIDE

Measuring downtime before it measures you

Record machine stops where they happen, give them reasons people can act on, and build availability, performance and quality from data you already trust.

Stockholm street and car with directional camera-motion blur.MANUFACTURING / SUNDIAL
Photography: Danial / Unsplash ↗ · Sundial motion treatment
THE PRACTICAL TAKEAWAY

Capture every stop at the workstation with a start, an end and a reason. Only then compute availability, performance and quality, and review the top reasons weekly with the people who can remove them.

Why downtime hides

Ask a plant manager how much time a machine lost last month and you will usually get a confident estimate. Ask for the list of stops behind it and the conversation gets shorter. Downtime hides because the people who see it are busy fixing it, and writing it down feels like paperwork that helps nobody.

The cost of not recording it is real but invisible. Late orders get blamed on sales promising too much, overtime gets blamed on absenteeism, and the same jam on the same machine keeps happening because nobody can show it happens every Tuesday. You cannot reduce what you do not record, and you cannot argue for a spare part or a maintenance window without evidence.

Record it where it happens

The record has to be made at the machine, at the time, by the person who saw it. Anything else turns into reconstruction at the end of the shift, which is where accuracy goes to die.

Sundial's Manufacturing module has a downtime entry for exactly this: the workstation, when the stop started, when it ended, and why. The operator or supervisor records it from a phone or a shared terminal near the line. The same workstation records the job cards that show what was being produced, so a stop is never just a gap; it sits next to the work it interrupted.

Keep the form short. Workstation, start, end, reason, and an optional note. Every extra field you add is another reason for a busy operator to skip it. If you want more detail on a particular class of stop, ask maintenance to add it when they investigate, not the operator when they are restarting the line.

Reasons that lead to action

A reason list is only useful if each reason points to someone who can fix it. "Machine problem" points to nobody. "Changeover", "waiting for material", "waiting for operator", "breakdown: electrical", "breakdown: mechanical", "quality hold" and "planned maintenance" each point to a different owner: planning, stores, the shift supervisor, maintenance, quality.

Resist the urge to build a list of forty reasons. People choose the first plausible option when the list is long. Start with eight to ten, watch which ones get used and what ends up in "other", and refine once you have a few weeks of entries. Separate planned stops (scheduled maintenance, planned changeovers, no orders) from unplanned ones, because the conversations about them are different.

From downtime to OEE, honestly

Overall equipment effectiveness (OEE) is the standard way to summarize how well a machine is used. It multiplies three ratios. Availability is run time divided by planned production time, so it drops with every unplanned stop. Performance compares actual output with what the machine should have produced in the time it ran, at its ideal cycle time. Quality is good units divided by total units.

Each ratio needs a different source, and each source has to be trustworthy before the number means anything. Planned production time comes from the workstation's working hours. Stops come from downtime entries. Output and run time come from job cards. Good versus rejected quantities come from the work order and any quality inspection. In Sundial these all live on the same workstation and work order, which is what makes the calculation possible without stitching three spreadsheets together.

Be careful with the headline number. OEE moves for many reasons at once, and a single percentage invites argument about whether it is good. The three components are more useful than the product: a machine with low availability needs maintenance and changeover work; low performance points to slow cycles or minor stops; low quality points to the process or the material.

Review it weekly, briefly

Downtime data earns its keep in a short weekly review. Look at the top reasons by lost time per workstation, pick one or two to attack, name an owner and a date, and check the next week whether the reason shrank. Do not try to fix everything; fix the biggest one and move to the next.

Feed breakdowns into maintenance planning. If the same machine stops for the same mechanical reason every few weeks, that is a preventive maintenance task waiting to be written. Sundial's Assets & Maintenance module keeps the maintenance schedule for the same equipment, so the stop history and the maintenance plan sit in one system rather than in two departments' files.

We are building Sundial so that the weekly summary does not have to be assembled by hand: you should be able to ask in plain language which workstation lost the most time last week and why, and get the answer with the entries behind it. That is the direction, and it depends entirely on the entries being made in the first place.