Machine data tells you that a machine stopped and for how long. It does not tell you why. Without the why, the data ends up in a chart that everybody looks at and nobody can act on.
Reason codes close that gap.
A short catalog
The catalog lists the reasons an operator can choose when a machine stops. The most common mistake is making it long. A list of sixty reasons is never used correctly, and the most popular choice becomes “other”.
Rules that work:
- Eight to twelve reasons on level one, the same in all areas of the plant.
- A definition for each: when it counts and when it does not. “Changeover: from the last good part of the old order to the first good part of the new one.”
- A threshold: only stops longer than three minutes need a reason. Shorter stops are counted as a group.
- An owner: one person changes the catalog, with a recorded decision.
In time, at the machine
The operator chooses the reason at the terminal within the shift, ideally when the machine restarts. Two touches are acceptable, five are not. In one vendor test, operators chose the system where entering a reason took two touches instead of five, even though it scored slightly lower on paper.
Track the share of stop minutes with a reason. Above 90 percent the data is usable. If “other” rises above 5 percent, the catalog or the training is wrong.
The loss Pareto
With reasons, one week of stops becomes a ranking. For twelve machines with 81,000 planned minutes per week:
| Reason | Minutes per week | Cumulative |
|---|---|---|
| Waiting for material | 4,620 | 22 percent |
| Changeover | 4,180 | 42 percent |
| Technical breakdown | 3,960 | 61 percent |
| Waiting for operator | 2,850 | 74 percent |
| Tool change | 2,210 | 85 percent |
Three reasons cause 60 percent of the lost time. The improvement work starts there.
From Pareto to action
A Pareto is useful only when the daily meeting acts on it. Every day the team looks at the top three losses of yesterday and asks: is there an action, who owns it, by when?
In the example, the largest loss was waiting for material. It was not a machine problem and not a software problem. A milk run every 60 minutes and a small supermarket at the hall entrance cut it by more than half within two months. Changeover, the second loss, went into a setup reduction workshop on the two weakest machines.
Keep the catalog alive
Review the catalog once per quarter with the team leaders. Merge reasons nobody uses, split one that hides two different problems, and keep the first level identical across all areas. Only then can the plant compare its losses and learn from the area that solved them first.