Evidence worth collecting before the condition changes
- Time-stamped stop reason and duration
- Line state before and after each stop
- Output by hour or shift
- Upstream/downstream starvation or blocking
- Manual interventions, changeovers, cleaning, quality holds, and material waits
Questions that improve the diagnosis
- Which event consumes the most total time, not only the longest single stop?
- Where does work accumulate or disappear?
- Is the constraint equipment, material, labour, quality, utility, or scheduling related?
- Can existing PLC/VFD/counter signals provide trustworthy state evidence?
What not to conclude too early
- Optimising the fastest machine instead of the constraint
- Calling all non-production time 'downtime' without categories
- Buying monitoring hardware before defining the decision