Chronic minor stops are short, repeating stoppages of a few minutes or less, a jam or a misfeed, that clear without maintenance but happen dozens of times a shift. Because each is trivial and none gets logged, they hide while stealing more capacity than the breakdowns everyone tracks.
The classic picture is a thirty-second jam that an operator clears with a flick of the wrist, forty times a shift. Nobody calls maintenance. Nobody writes it down. Every stop feels like nothing. But forty stops of thirty seconds is twenty minutes of downtime plus the ramp-back-to-rate after each one, and the same jam tomorrow, and the day after. This is idling and minor stops, the third of the classic six big losses and it is the loss most plants underestimate the most, precisely because it never triggers an alarm. This post explains why it is a design problem rather than an operator problem, and how to attack it.
Why are chronic minor stops so easy to miss?
They fall below every threshold that would catch them. Downtime logging usually starts at a minimum duration, five minutes is common, so a run of ninety-second stops never enters the record at all. The operator who clears them is not idle during the stop, they are working, so it does not feel like downtime to anyone watching. And no single instance is worth reporting, so the reporting instinct never fires. The loss is real, but every mechanism you would use to see it has been tuned to ignore it.
The result is a systematic blind spot. A line can post decent availability, because availability only counts the stops long enough to log, while its performance term quietly bleeds. Chronic minor stops mostly land in the performance term of OEE calculation not availability, because the line technically keeps running between them; it just never holds rate. That is why a plant can have "good" downtime numbers and a mediocre OEE it cannot explain. The gap is full of stops too short to see.
What actually causes chronic minor stops?
Small, tolerated variation in the things that feed and move the product: a package a millimeter out of fold, a label web with inconsistent tension, a photo-eye positioned where dust trips it, an infeed that surges instead of metering. None of these is a broken part. Each is a condition the equipment was never quite set up to handle, so it stumbles, the operator recovers it, and the condition remains for the next cycle to stumble on again. Chronic minor stops are the equipment telling you, dozens of times a shift, that some input is out of the range it can absorb.
This is why they are stubborn. A breakdown has a failed component you can replace; a minor stop has a marginal condition that no single repair fixes. The jam clears, the line runs, and nothing has changed, so it jams again. Chasing the symptom, clear, run, clear, run, can go on for years because the loop never touches the cause. Breaking it requires treating the recurring stop as a defect to be engineered out, which is root cause analysis territory, not a housekeeping task.
Why is it a design problem, not an operator problem?
Because the same stop happens no matter who runs the line. If your best operator and your newest operator both fight the identical thirty-second jam forty times a shift, the jam is not a skill gap; it is built into the process. Blaming the operator is not only unfair, it is a diagnostic error: it points the fix at training when the fix is a tooling, sensor, or material change. The operator's fast recovery is actually masking the problem, keeping the line limping so the underlying condition never gets the attention a full stop would force.
There is a trap here worth naming. An operator who is very good at clearing jams makes the loss invisible, which removes the pressure to fix it. The better the recovery, the more permanent the defect becomes, because it never hurts enough to escalate. That is why plants that only reward "keeping the line running" can carry the same minor stop for a decade. Making the loss visible, counting it, is the precondition for anyone deciding it is worth engineering out. The countermeasure is usually poka-yoke: change the tooling or the sensing so the stop cannot occur, rather than training people to recover from it faster.
How do you attack chronic minor stops?
You cannot fix what you cannot see, so the sequence starts with making the invisible loss visible, then engineering out the top offenders one at a time:
- Make the stops visible. Drop the logging threshold or add automatic detection so stops under five minutes get counted. If capture is manual, run a focused observation: one person, one line, one shift, tallying every stop and its cause.
- Pareto the stop reasons. A handful of causes will account for most of the count. Rank them with a Pareto chart by frequency, because with minor stops the count matters as much as the minutes.
- Watch the top cause at the machine. Stand at the line and observe the number-one stop until you see the physical condition that triggers it. Minor stops rarely reveal themselves in a report; they reveal themselves at the machine.
- Ask why it recurs, not why it happened. Run the five whys on the recurring condition, not the single instance. The goal is the marginal input, the tension, the alignment, the tolerance, that the equipment cannot absorb.
- Engineer the stop out. Change the tooling, reposition or shield the sensor, add a metering device, tighten the incoming spec. Mistake-proof the condition so recovery is unnecessary because the stop no longer happens.
- Confirm with the count, not the feel. Verify the fix by watching the stop count for that cause fall, then move to the next line on the Pareto. Feel is unreliable here because the loss was always below the feel threshold.
- Fold the wins into autonomous care. Make the new setup, cleaning, or inspection part of routine operator care so the condition does not creep back, tying it to your total productive maintenance program.
How do you measure minor stops when they are too short to log?
Either lower the human threshold with a focused count, or remove the human from the loop with automatic detection. The focused count is cheap and immediate: one observer, one line, one shift, a tally sheet with a row per cause. It will not run forever, but it produces the Pareto you need to start. The durable answer is machine-level detection that timestamps every stop regardless of length, so the short ones stop escaping the record. This is where reason-coded machine downtime capture pays off, because it counts the stops a five-minute threshold would erase.
How much do chronic minor stops actually cost?
More than their length suggests, because each stop carries a recovery tail. A thirty-second jam is rarely a thirty-second loss: the line coasts down, the operator clears it, and then the line ramps back to rate, so the real cost per stop can be double or triple the stop itself. Multiply that by frequency and a loss everyone dismissed as trivial becomes one of the biggest single items on the line. This is exactly why idling and minor stops is the loss the framework flags as most underestimated.
The standards and data back up treating it seriously. Idling and minor stops is one of the six equipment losses formalized in the international manufacturing KPI standard, ISO 22400-2, which defines OEE and its supporting metrics so short stoppages are not quietly excluded (ISO 22400-2:2014). And with U.S. manufacturing capacity utilization running in the mid-70s percent range, most recently about 75.7 percent (Federal Reserve, G.17), the performance that chronic minor stops erode is capacity the plant already owns and pays for. Recovering it costs engineering time, not new equipment.
How do minor stops connect to the rest of the line?
They ripple, especially at the constraint. A minor stop on a non-bottleneck machine may be absorbed by buffers, but the same stop on the constraint is throughput gone for good, because the constraint sets the pace for the whole line and lost time there cannot be recovered downstream. That is why the first place to hunt minor stops is the bottleneck: a thirty-second jam there is thirty seconds off the entire plant's output. Reducing it lifts throughput in a way the same fix on a non-constraint never will.
Getting there requires numbers arriving fast enough to act on, which is where the loss usually breaks down. Paper logs never catch sub-threshold stops, so the loss stays invisible and unfunded. Real-time capture at the station counts every stop as it happens, surfaces the Pareto automatically, and shows the count fall when a fix works, all without waiting for a monthly report. That move from paper logging to live capture is what CLS built across its shops (see the CLS case study), and it is what turns chronic minor stops from a tolerated nuisance into a tracked, shrinking line item. You can size the OEE upside of recovering that lost performance with the OEE calculator. No rip-and-replace, just the small stops finally counted.