This article is for production managers, plant managers, continuous improvement leads and team leaders who work with lean and run a daily huddle on KPIs. It covers three losses that cost capacity without showing clearly on the board: rework, waiting time and micro-stops. You will see why they often stay out of sight, which KPI tracks each one, how to measure them machine by machine instead of estimating them, and how to turn them into hours of capacity the daily huddle can act on.
- Rework, waiting time and micro-stops are small each time and rarely written down.
- Each loss has its own KPI and its own place in OEE.
- Converted into hours per week, they become comparable with big breakdowns.
- The daily huddle only acts on them well once they are measured at the machine.
What Lean.nl says about KPIs in the daily huddle
In the Lean.nl knowledge base, Anend Harkhoe explains in the Dutch article Lean Manufacturing uitgelegd how lean works in production: flow, stability, quality at the source, visual management and waste reduction, with practices such as 5S, Value Stream Mapping, SMED and TPM. In the part on data in the daily huddle (the Dutch “dagstart”), the article groups KPIs around delivery reliability, flow, stability and quality. Under quality, next to First Time Right, one short sentence sums up the topic of this article: “Rework rate maakt herstelwerk zichtbaar dat capaciteit opslurpt.” (“Rework rate makes visible the rework that eats up capacity.”)
Lean.nl links lead time to the waiting and congestion inside the process, and places downtime minutes and OEE under stability. For each KPI, the article asks four questions: what deviates and where, what is the main cause today, what action do we take now, and who does it and when is the result checked. This article builds on that for one specific point: the losses that eat up capacity while the machine looks like it is running or available on the board, and how to measure them on the shop floor.
Three losses that cost capacity without counting as breakdowns
A one hour breakdown gets noticed. Everyone saw it, it is in the log and it comes up in the huddle. Rework, waiting time and micro-stops behave differently: each one is small, they are spread over the day and they are rarely written down. That is why they seldom top the Pareto chart, even when together they cost more hours than the big breakdown.
Rework. A part that has to be processed again uses the machine a second time, or ties up a repair station and an operator. In the end the part is good, so anyone who only counts scrap does not see the loss. In OEE it sits in the quality factor, but the time the repair takes elsewhere is not counted there.
Waiting time. The machine is technically fine but stands still because material, an operator, a release or the upstream process is missing. On many boards this does not count as a breakdown, sometimes not even as downtime, because the cause is organisational. As long as the time was planned as production time, in OEE it is an availability loss.
Micro-stops. Stops of a few seconds to a few minutes that the operator clears alone: a jammed product, a dirty sensor, a feeder that hesitates. They are almost never logged by hand. In OEE they then end up in the performance factor, as a lower speed with no cause attached.
Which KPI for each loss, and how to measure it on the shop floor
| Loss | KPI | Shop floor measurement | Place in OEE | Pitfall |
|---|---|---|---|---|
| Rework | Rework rate: share of parts processed again, plus repair time | Count at the source, per machine where the defect started, with the reason | Quality, plus extra machine time when the repair runs on the same machine | Counting scrap only: reworked parts vanish into good output |
| Waiting time | Minutes of downtime with a waiting cause (material, operator, release, upstream process) | Automatically detected stop, cause assigned by the operator at the machine | Availability, when the time was planned | Keeping waiting out of the measurement as not our downtime |
| Micro-stops | Number of short stops per hour and total minutes per shift | Automatic detection of every stop, including stops of a few seconds | Performance or availability, depending on the threshold you choose | A logging threshold of several minutes that makes them invisible |
From minutes to hours of capacity: the method in five steps
- Pick the bottleneck machine. Capacity recovered there turns into extra output. On another machine it mostly turns into more waiting further down the process.
- Measure a full week without filters. Every stop, short ones included, with a cause per stop, plus the number of reworked parts per day with the reason.
- Convert each loss into hours. Waiting time and micro-stops are already in minutes. For rework, multiply the number of reworked parts by the repair time per part, and add the machine time if the repair runs on the same machine.
- Put the three next to the big breakdowns. In hours per week they become comparable, and the order of attack becomes clear.
- Take it to the daily huddle. For each KPI, the four questions from the Lean.nl article: what deviates, what is the main cause, which action, who carries it out and when is it checked. The following week, measure the same KPI again.
An example for illustration, not a measurement: suppose a machine has forty micro-stops per shift, thirty seconds each on average. That is twenty minutes per shift, and one hour a day over three shifts, while no single stop was long enough to be written down. Add waiting for material at every shift start and a handful of parts that go back into the machine each day, and the huddle sees a capacity loss the breakdown board never showed.
Why estimates are not enough here
Manual logging works reasonably well for long breakdowns. It falls short precisely on these three losses. An operator clears a micro-stop instead of writing it down, waiting time is filled in afterwards as other, and rework happens at the end of the shift somewhere else. Lean.nl points out that measuring only helps when the data is understandable and directly linked to the work. For these losses that means measured at the machine, at the moment it happens, with a cause the operator recognises.
Two points keep the discussion healthy. Compare causes, not people: a shift with more micro-stops may simply be running a harder product. And let the operator assign the cause at the machine, so the data stays the team’s own and does not feel like surveillance.
Where TEEPTRAK fits
TEEPTRAK measures downtime and micro-stops automatically: a module in the electrical cabinet and a sensor on the line, about one hour of installation per machine, with no connection to the PLC, and a first OEE within 48 hours. Every stop is detected, short ones included, and the operator assigns the cause at the machine, so waiting for material or for a release gets its own cause. Rework still needs a count at the source, which you put next to the measured stops. TEEPTRAK is not an MES and does not replace the lean method: it provides the measured data the daily huddle relies on. It is used in more than 700 plants in more than 30 countries, with an average gain of 10 to 15 OEE points.
Further reading
- Lean Manufacturing uitgelegd (Lean.nl, Anend Harkhoe, in Dutch)
- Presenting production losses to management: from percentages to hours and capacity
- Measuring waiting waste: machine downtime in a lean waste analysis
- How to Calculate the Overall Equipment Effectiveness (OEE)? (dmaic.com, a worked explanation of the OEE calculation)
FAQ
Does waiting for material count in OEE? Yes, if the time was planned as production time: it is then an availability loss, even though the cause lies outside the machine. Give it its own cause so production and logistics can discuss it together.
From what duration is a stop a micro-stop? There is no fixed standard, each plant sets its own threshold. What matters more is that every stop is measured, so the threshold shapes the analysis rather than what happens to get written down.
How do I measure rework when it happens somewhere else? Count the reworked parts against the machine where the defect started, with the reason, and record the repair time separately. That way the loss lands at its source and not at the repair station.
0 Comments