Why job change is the highest-leverage operational lever
Job change — the changeover from one container SKU to another — is the single largest source of controllable hot-end downtime in most container glass plants. Cross-shift variance on the same SKU changeover is typically 30–60%. A 1% efficiency lift across an annualised plant network is worth millions in EBITDA. Yet most plants still rely on paper checklists, tribal knowledge, and a vendor sales deck to manage what is genuinely the highest-leverage operational discipline they have access to.
This guide is the operator's complete reference to job change in container glass — the lifecycle, the tooling, the KPIs, the failure modes, and how to install the discipline in your plant.
The 9-stage Job Change Lifecycle
Job change is not an event — it is a lifecycle that begins 72 hours before the line stop and runs for 24 hours after restart. The 9 stages:
- 01 Planning (T-72h) — SKU selection confirmed, mould inventory verified, crew briefed, target timings agreed
- 02 Pre-job change SOP (T-24h) — staging carts pre-loaded, mould preheat scheduled, forehearth setpoints pre-calculated
- 03 Mould preheat & seasoning (T-2h) — incoming moulds preheated to target; failure here causes the bulk of first-ware defects
- 04 Line stop / hot end clear-down (T-0) — controlled ramp-down, last saleable ware captured, equipment isolated
- 05 Equipment changeover (T+0 to T+45min) — mould swap, plunger swap, neck ring change, baffle change, takeout adjustment
- 06 First ware (T+45min to T+90min) — restart begins; first 200 ware are typically rejected; first-ware quality KPI starts here
- 07 Stabilisation (T+90min to T+4h) — pack rate climbs to target; defect modes settle out
- 08 Sign-off + KPI capture (T+4h) — KPIs locked: changeover time, first-ware quality, percent-pack, defect mix
- 09 Post-mortem (T+24h) — what worked, what didn't, SKU Library updated
Target timings, KPIs and exit criteria per stage
Best-in-class plants run target timings as follows. These are achievable; we have installed them at multiple plants:
- Stage 05 (equipment changeover): 30–45 minutes
- Stage 06 (first ware): first saleable ware within 60 minutes of restart
- Stage 07 (stabilisation): >90% pack rate within 240 minutes
- Total wall-clock changeover (line down to first saleable): under 6 hours
- First-hour yield (saleable / produced in first 60 min): >85%
- Cross-shift variance on changeover time: <15%
What goes wrong at each stage
Failure modes are remarkably consistent across plants. The most common:
- Stage 01 Planning — wrong mould set in inventory, gap in mould availability, scheduling conflict with maintenance
- Stage 03 Preheat — variable preheat time across crews, preheat oven capacity constraints, mould temperature not measured
- Stage 05 Changeover — staging cart not pre-loaded, plunger setup discrepancies vs SKU spec, role ambiguity at swap
- Stage 06 First ware — cold mould seasoning, forehearth not at SKU setpoint, swab programme not switched
- Stage 07 Stabilisation — defect mode persists, root cause not surfaced before crew handover
- Stage 09 Post-mortem — does not happen, or happens without data
Paper checklist vs systemised job change
Paper checklists are the legacy approach to job change. They have three structural weaknesses: they are not auditable, they do not adapt across crews and shifts, and they do not feed back to a learning loop. The systemised replacement — combining the Job Change Tool with the Lifecycle discipline — fixes all three.
SKU Library — locking in best-known state
Every job change involves dozens of decisions that should be deterministic for a given SKU: gob weight, forehearth setpoints, mould cooling configuration, swab programme, plunger setup, takeout timing, hot-end coating dose, lehr profile, target changeover time. In paper systems these are re-discovered each run. In the SKU Library, they are locked in, versioned, and updated only by deliberate change control.
Live Execution — guided, timed, checked
During the changeover, every role (mould changer, forehearth, IS operator, QA, hot-end superintendent) sees the next step assigned to them on a tablet, with target timing, quality check, and dependencies on other roles. Steps are completed with photo or sign-off evidence. Deviation from target is surfaced live, not at end-of-shift.
KPI Tracking — closing the loop
Changeover time, first-ware quality, percent-pack, defect mix and OEE recovery are tracked live, target vs actual. Every changeover writes to the SKU Library — either confirming the standard or proposing an update via the change control process. The system gets smarter every job.
Worked example — sub-six-hour changeover
Anonymised example from a recent engagement. European spirits plant, 8 SKUs, 14-week installation. Pre-engagement: average changeover 14h 20min, first-hour yield 62%. Post: average 6h 40min, first-hour yield 91%. EBITDA recovery: $3.4M annualised. Changeovers per year × time saved × pack rate × selling price × margin — the maths gets large quickly at any plant doing 40+ changeovers per year per line.
How to install this in your plant
The installation pattern is consistent: 2 weeks of video study and analysis, 2 weeks of standard drafting with the crews, 6–10 weeks of piloted execution and iteration. By week 14, the plant has a written standard, a dashboard, and at least ten changeovers at the new target. The pilot line then teaches the rest of the plant; rollout pace is set by capability transfer, not calendar.
Lean Glass installs both the discipline (Job Change Lifecycle) and the tooling (Job Change Tool) — under one engagement or two. Most clients run them together for the closed-loop effect.