Death by a thousand interruptions

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 interruptions, short halts of seconds at a time (what standard OEE calls minor stops or micro-stops), 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.

Seeing when the interruptions happened did the real work. It told the team where to look. The causes were then found at the line, by the plant's own engineers. 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.