Learn Automation with MINATA #58: OEE Data — Define the States Before You Calculate a Percentage
OEE Data: Define the States Before You Calculate a Percentage
Overall equipment effectiveness is an attractive percentage, but it is only trustworthy when the machine states have been defined first. Label every stoppage "machine fault" and you get a wrong OEE, followed by wrong decisions.
Three components
OEE is availability multiplied by performance multiplied by quality. Availability measures the time the machine was able to run against the plan; performance measures actual speed against design speed; quality measures the proportion of good product. A low OEE does not say why on its own — only separating the three components tells you whether to fix availability, speed or quality.
Downtime reason codes
Not every stoppage is a machine fault. You need reason codes and a rule for telling them apart:
| Group | Example | Counts as |
|---|
| Setup and changeover | Tool change, adjustment | Planned; not a machine fault |
| Starved | No incoming material | External to the machine |
| Blocked | The downstream machine is full | External to the machine |
| Fault | A technical failure | Reduces availability |
| Maintenance | Planned or unplanned | According to type |
Without reason codes, every stoppage piles into "fault", and maintenance gets blamed for a material shortage.
Trustworthy data sources
OEE data needs timestamps, machine state and a reliable counter from the PLC. People should only enter what genuinely requires judgement — the reason for one long stoppage. Count product from a real confirmation signal, not from a conveyor command. Automatic data reduces both the error and the arguing in the improvement meeting.
A worked engineering situation
The PLC classifies Run, Starved, Blocked and Fault from sensors and logic; the HMI asks the operator to choose a reason for long stoppages; the dashboard aggregates OEE by shift and by product code. Because Starved and Blocked are separated from Fault, the improvement team can see that most of the loss comes from waiting for material rather than from breakdowns — and puts its effort in the right place.
The state definitions, the reason codes and the calculation all follow the plant's own convention and its production process.
Common mistakes
- Labelling every stoppage a machine fault.
- Not separating availability, performance and quality.
- Counting product from a command rather than a confirmation signal.
- No reason codes, so nobody knows where the loss comes from.
- Entering everything by hand, producing biased data and disputes.
OEE data checklist
- [ ] The machine states are defined before any percentage is calculated.
- [ ] Reason codes distinguish setup, starved, blocked, fault and maintenance.
- [ ] Timestamps, states and counters come automatically from the PLC.
- [ ] Product is counted from a real confirmation signal.
- [ ] Availability, performance and quality can each be separated and acted on.
Defining the states properly is the foundation of an OEE you can use. #59 turns to IIoT connectivity, starting from the data question and the security boundary.
Read more automation knowledge at MINATA: https://minatavn.com/en/blog/industrial-automation
Previous — #57: Energy metering in the control panel: https://minatavn.com/en/blog/automation-57-panel-energy-metering
Next — #59: IIoT connectivity: https://minatavn.com/en/blog/automation-59-iiot-connectivity
View all MINATA technical articles