A packing site had an OEE of 26.5% and no obvious villain. There was no big breakdown to point at. The losses were 218,000 tiny stops, each too short to notice, all dumped into a single uncategorised bucket that hid the real story.
The line's OEE sat at 26.5%, low enough to be alarming, and yet there was no obvious cause. The maintenance log was quiet. There had been no major breakdown, no long outage, nothing dramatic enough to explain why the line was producing at barely a quarter of its potential. By every account the line was "running fine". It simply wasn't producing.
When a line is losing that much time with nothing to point at, the losses are almost always small and frequent rather than large and rare. The problem is that the tracking in place could not see small and frequent.
Augos captured the line's stops at a much finer resolution than the site's existing system. The result was stark: over the measured period the line had suffered 218,000 individual micro-stops, short halts of seconds at a time, each one previously invisible because it was too brief for the old tracking to register. Every one of them had been swept into a single uncategorised downtime bucket.
A quarter-million stops collapsed into one line item looks like an unsolvable mess. Split out by cause, it becomes a short list of specific, fixable problems, each responsible for tens of thousands of the stops.
Categorising the micro-stops did the real work. Instead of 218,000 problems there were a handful of root causes, a feed that jammed, a sensor that tripped, a station that choked, each accountable for a huge share of the total. That turned an intimidating, formless loss into an ordinary engineering backlog: fix the top cause and stops disappear in their tens of thousands.
The wider point is about measurement, not machinery. A line can be quietly losing three-quarters of its capacity to stops no one is recording. If the data can't see a two-second halt, it can't see the biggest loss on the floor, and the OEE number stays a mystery.
A very short stoppage, often seconds, where the line halts and restarts. Individually they are trivial; at scale they can be the single largest loss on a line, and coarse tracking never records them.
The existing tracking recorded downtime at a resolution too coarse to see short stops, so anything that didn't fit a named category was swept into a single 'other' or uncategorised total.
You don't fix them one by one. Once they are categorised you find the handful of causes behind the vast majority, and fixing those causes removes stops in their thousands.
Book a demo with one of our manufacturing experts.