Why SMED translates (and where it doesn't)
Single-Minute Exchange of Die (SMED) was developed by Shigeo Shingo in automotive stamping. Direct application to container glass IS machine job change does not survive first contact with a forming machine — but the core disciplines translate:
- Separate internal from external work
- Convert internal to external where possible
- Streamline whatever remains internal
- Standardise, then iterate
The constraint that doesn't translate: a stamping press can be re-tooled in minutes; a hot-end IS machine cannot. Mould preheat alone is multiple hours. So 'sub-minute' is not the target — sub-six-hour is, and at that scale the gains are still enormous.
Internal vs external work in container glass changeover
Internal work happens while the line is down. External work happens while the line is still running (or after restart). Common shifts from internal to external in container glass:
- Mould preheat — convert to fully external by pre-heating in dedicated ovens
- Plunger setup — pre-set on bench, swap as a unit
- Forehearth setpoint calculation — pre-calculate, not done during the changeover
- Swab programme update — staged in the SKU Library
- Crew briefing — done at T-24h, not at T-0
The video study
Every SMED implementation in container glass starts with a video study. Film 2–3 changeovers across crews and shifts. Classify every minute of work as internal, external, or waste. The findings will name themselves — typically 30–40% of currently-internal work can move external with no equipment investment.
The choreography
A six-hour changeover is not one team racing. It is three to five roles running in parallel, handing off at specific moments, with a shared timeline everyone can see. We map it like a recipe — T-60 to T-0 (running, external prep), T0 to T+30 (shutdown), T+30 to T+180 (parallel execution), T+180 to T+300 (restart ramp), T+300 to T+360 (stabilisation handover).
Pre-staging carts and tool boards
Modest investment in dedicated staging carts and tool boards typically pays back within 2–3 changeovers. The principle: no operator should be looking for a tool during the line-down window. Everything they need is at the workstation, in a known position, on a board with a shadow.
Standard work that survives the bad shift
Documented role cards laminated at each station — two or three bullets per role per phase. The video record of the new standard is more durable than the document. Crews who film themselves doing the new standard well rarely revert.
Common SMED mistakes in container glass
- Treating SMED as a tooling project — it is a discipline project
- Optimising changeover time without measuring first-ware quality — fast changeover with bad first ware is not a win
- Rolling out without a pilot — every line has its own quirks
- Documenting standards in PDF that nobody reads — use tablet-based guided execution instead