The six big losses
The classification comes from TPM (Total Productive Maintenance) and is the standard way to explain a low OEE. Every minute lost from planned time ends up in one of the six categories, and each category weighs on only one of the three OEE factors.
| Loss | Examples | OEE factor |
|---|---|---|
| Breakdowns | A component failure, a maintenance intervention, a stop on a machine alarm | Availability |
| Setup and adjustments | Die or format change, tooling, warm-up, adjustments at startup | Availability |
| Micro-stops and idling | Jams, a misfed part in the feeder, a dirty sensor, waiting a few seconds for an operator | Performance |
| Reduced speed | A cycle slower than the ideal cycle, speed lowered to avoid problems, wear | Performance |
| Process defects | Nonconforming parts and rework during steady-state production | Quality |
| Startup rejects | Parts scrapped after a startup or a changeover, until the process is stable | Quality |
The table has a practical consequence: to know which factor a loss weighs on you need to know what kind of stop it is, and for the first two rows that takes a reason code. For micro-stops, on the other hand, the reason is almost always missing: they are too short for anyone to pick one.
What micro-stops are
A micro-stop is a short stop that is cleared without maintenance: the operator removes a jammed part, cleans a photocell, resets an alarm and the machine starts again. It happens dozens of times per shift, almost always with the same causes, and that is why it gets mistaken for normal running.
In OEE, micro-stops belong to performance, not availability: during a micro-stop the machine is considered running, but it is not producing. That is why a machine with high availability and low performance should be checked first of all for micro-stops and reduced speed.
Threshold: when a stop is micro
There is no standard threshold. In practice a stop under 5 minutes is treated as a micro-stop, and many companies choose lower values, from 1 to 3 minutes, for fast machines. What matters is that the threshold is written down, the same for all machines of the same type, and does not change from one month to the next.
- Below the threshold: a micro-stop, counted and summed, with no reason code required. It weighs on performance.
- Above the threshold: a real stop, with a reason chosen by the operator or read from the PLC. It weighs on availability.
- A cycle longer than normal but without a stop is reduced speed, not a micro-stop: you need to tell them apart if you want to understand the cause.
Moving the threshold does not change OEE: it moves minutes between availability and performance. If OEE changes when the threshold changes, the calculation is losing part of the time or counting it twice.
How to count them from the PLC
Micro-stops are not counted by hand: there are too many and they are too short. They are derived from the signals the PLC already has, read continuously.
- Machine state (in cycle, stopped, in alarm): every transition from in cycle to stopped opens a stop, the return to in cycle closes it.
- Part or cycle counter: if the counter does not advance for more than a multiple of the ideal cycle time, for example twice, the machine is not producing even if the state says otherwise.
- Alarm or stop code, where it exists: it gives a cause even to short stops and makes a Pareto of micro-stops possible.
- Frequent reading: with slow sampling, stops of a few seconds disappear. It is better to read states on change, for example with an OPC UA subscription, or with one-second sampling.
With Muvia Industrial the edge node reads states and counters from Siemens S7, OPC UA and Modbus TCP PLCs with its drivers and keeps them in a local buffer if the network goes down. Your team writes the calculation that separates stops from micro-stops in a Python function, once for all machines; Muvia has no packaged OEE module.
Example: a press with 40 micro-stops per shift
A press has 450 planned minutes per shift and an ideal cycle time of 30 seconds. The PLC signals show 40 stops under the 5-minute threshold, with an average duration of 1.5 minutes, plus 45 minutes of real stops.
| Quantity | Calculation | Result |
|---|---|---|
| Time lost to micro-stops | 40 × 1.5 min | 60 min |
| Parts not produced | 60 min ÷ 0.5 min per part | 120 parts |
| Impact on performance | 60 ÷ 405 min of run time | 14.8 points |
| Impact on OEE | 60 ÷ 450 planned min | 13.3 points |
The next step is to understand when they happen. A day-by-hour map of micro-stops often shows a clear pattern: rows that light up at every shift start after a die change, or an afternoon hour when the feeder jams. This is the path of the use case OEE and losses on stamping and welding lines, where micro-stops by hour and the Pareto of causes tell you where to act.
Frequently asked questions
Breakdowns and setup (they weigh on availability), micro-stops and reduced speed (they weigh on performance), process defects and startup rejects (they weigh on quality).
There is no single rule: a stop under 5 minutes that is cleared without maintenance is usually considered a micro-stop, and many companies use 1–3 minutes for fast machines. The threshold should be written down and kept the same for machines of the same type.
On performance. During a micro-stop the machine is considered running but is not producing, so the lost time reduces performance, not availability.
From the PLC signals: the machine state and the part counter, read continuously. Every interval in which the machine is not in cycle, or the counter does not advance for more than a multiple of the ideal cycle time, is a stop; those under the threshold are micro-stops.