Define the line boundary and saleable-pack count first. Calculate availability, performance and quality separately, then compare stop, speed and reject losses by product and format. The station with the largest queue is not automatically the true bottleneck.
Why this matters
OEE is useful when everyone agrees what planned production time, ideal speed, total count and good count mean. On integrated filling and packaging lines, short stops, downstream blocking, upstream starvation, replenishment and changeovers can distort the headline figure. Use OEE to rank losses, not as a target that encourages teams to hide events.
Decision table
| Measure | Definition | Common error |
|---|---|---|
| Availability | Run time ÷ planned production time | Inconsistent treatment of planned stops |
| Performance | Ideal cycle time × total count ÷ run time | Using one catalogue speed for every SKU |
| Quality | Good count ÷ total count | Counting rework or quarantine as good output |
| OEE | Availability × performance × quality | Comparing results created from different boundaries |
| Constraint test | Good output while blocked and starved states are recorded | Assuming the most visible queue identifies the cause |
Information to collect first
- Choose one representative product and format for the baseline.
- State where the line begins and where saleable output is counted.
- Record every stop with initiating reason, duration and recovery time.
- Track micro-stops and reduced speed as well as long breakdowns.
- Separate blocked time from starved time at each interface.
- Repeat the measurement after one controlled improvement.
Recommended method
- 1
Confirm stable materials and approved settings.
- 2
Rank losses by lost good packs rather than event count.
- 3
Protect the suspected constraint’s infeed and discharge for a controlled trial.
- 4
Convert recoverable minutes into realistic additional output.
- 5
Use the same definitions after the change.
Common mistakes to avoid
- Using OEE as a league table between teams.
- Changing several settings before the loss pattern is understood.
- Using theoretical speed where the pack has never run stably.
- Ignoring changeovers and replenishment because they are planned.
- Adding accumulation without proving short downstream stops are the cause.
Questions people also ask
What is the best OEE boundary?
Use the boundary that represents the commercial result being managed, normally from an agreed infeed condition to saleable output.
Can individual machine OEE values be added together?
No. Machine results help diagnose loss, but the integrated line result depends on the flow of good product through the full boundary.
Should changeovers be included?
Use a documented rule and keep it consistent. Track changeover duration separately even when it is removed from planned production time.
How does OEE support an investment case?
Convert proven recoverable time into good packs, contribution, labour and waste impact using conservative assumptions.
Continue to the appropriate specialist route
Use the links below after defining the product, pack, output and acceptance criteria.