Automation feasibility is weakest when the current job is summarized as a single ideal cycle. A useful record includes what enters the station, what changes, what interrupts work, what an experienced operator notices, and what proves completion.
1. Observe the real cycle
Record several representative cycles across products, operators, shifts, and upstream conditions. Separate active work from waiting, walking, checking, rework, replenishment, and recovery.
- Part families, presentation, orientation, and variability
- Tools, fixtures, consumables, utilities, and interfaces
- Normal sequence and acceptable operator variation
- Quality checks and evidence retained today
A cycle target without its measurement conditions is not an acceptance criterion.
2. Name the exceptions
Ask what makes the experienced operator pause. Missing parts, ambiguous orientation, damaged packaging, a full downstream buffer, and equipment not ready may define more of the solution than the nominal motion.
3. Define completion evidence
State what must be true before the station releases a part. Avoid vague labels such as “good quality.” Use observable criteria and name how a reject is contained, confirmed, and recovered.
| Question | Evidence | Owner |
|---|---|---|
| Correct part? | Readable identity and approved recipe | Process engineering |
| Correct completion? | Defined measurement or machine result | Quality |
| Safe release? | Interfaces ready and downstream capacity | Controls |
4. Plan ownership
Decide who owns samples, upstream variation, acceptance, operator training, source code, backups, spares, open risks, and future product changes. A technically successful cell can still fail operationally when ownership is left implicit.