IS Machine Changeover Discipline, Section by Section

It is 02:15 on a Thursday and section 6 is back down for the third time since midnight. The mould change on this eight-section NNPB line went eleven hours, and the night-shift operator is standing at the control panel trying to remember what plunger timing offset the previous crew used on this article. Nobody wrote it down. Nobody ever does.

That scenario plays out on container glass IS machine lines from the GCC to the Gulf Coast more often than any plant manager wants to admit. The changeover itself is rarely the failure point. The failure is that every operator runs it differently, every section comes up on its own schedule, and there is no shared reference for what stable means before someone declares the job done.

Blank mould preheat sets your time-to-quality before the first gob lands

Most of what goes wrong in the first 30 minutes after a job change is thermal. Not mechanical. Not a forming fault.

Blank moulds must reach 400–480°C core temperature, measured at the mould body midpoint with a contact pyrometer (not estimated from oven cycle time, which is what most mould prep teams will tell you is close enough). Surface temperature below 350°C produces chill-wrinkle, mould-seam grooving, and surface crystallisation. Rejection rates for an under-heated section run 20–35 percentage points above steady-state for the first 15–25 minutes of production. That is not a quality problem from the IS machine. That is a preparation failure.

The preheat protocol is the cheapest intervention available on any container glass IS machine line. A dedicated mould oven, a preheat cycle tied to the changeover timeline, a contact pyrometer check before release: none of that requires capital expenditure. What it requires is that someone owns the step. The hot-end superintendent should be signing off that every section's blank mould package has hit the 440°C floor before gob delivery starts. On most lines I've audited, that check is visual and the sign-off is verbal. That is why the first 20 minutes on a new article are so consistently bad.

In 2018 I was working with an operator on a five-section line in the GCC running a 280 g wine bottle format. They were losing an average of 18% pack-to-melt in the first hour after every job change. The root cause wasn't the forming setup. Blank moulds were going in between 310°C and 370°C depending on which mould prep technician handled them that shift. Once preheat was locked to a minimum 440°C verified pyrometer reading before release, time-to-quality dropped from 55 minutes per section to under 35. No new equipment. One owned procedure.

Section timing drift is a controller conversation, not a floor estimate

NNPB plunger contact time has a target window of 0.08–0.12 seconds. Post-changeover thermal re-equilibration of the gob feeder spout and plunger mechanism routinely produces timing drift of +0.025–0.040 seconds. That looks small. What it produces in the blank mould is a thick-heel, thin-shoulder parison geometry that manifests at the cold end as thermal-shock checks and pressure-test failures. Those defects get logged as forming problems. They are timing re-entry problems on the electronic section controller.

Twenty-three minutes. That is the window in which timing drift on an NNPB section either gets corrected or compounds into a defect pattern that takes another hour to unravel downstream.

The hot end tells you in real time. The cold end tells you 40 minutes later. Most plants wait for the cold end.

The fix requires someone at the Emhart AIS controller (or a Bucher IS unit if that is what the line runs) during the first 10–15 production cycles after restart, running the section timing sequence from the controller. Not adjusting gob weight. Not chasing cooling air. The operator who doesn't know what timing offset was entered on the previous article will reach for the nearest visible variable. That is how a timing problem turns into an hour-long gob weight investigation with no resolution.

And this is where shift handover compounds the damage. The operator who set up the section leaves at 06:00. The incoming crew inherits a section that is drifting and has no record of what timing parameters were last entered on the Emhart AIS. The 06:00 handover misses the overnight section-timing log on most IS machine lines I audit. The new operator adjusts gob weight. The defect pattern gets worse. The section comes down again. Nobody connects it to the timing entry that happened two hours before shift change.

Cooling air balance follows the same pattern. IS machine section cooling circuits typically run at 0.8–1.5 bar inlet pressure with manifold needle valves set per article geometry. After a changeover, those valves need re-balancing for the new format. Imbalanced cooling in the first one to two hours post-changeover produces asymmetric wall-thickness distribution across the container circumference, generating bird-swing precursors and eccentricity defects that fail pressure test. These land at cold-end inspection looking like IS machine forming faults. Most of the time they are cooling variables that were never re-set after the changeover.

Gob weight variance and the 30-minute recovery window

Steady-state gob weight tolerance runs ±1.0–1.5 g. Post-changeover thermal re-equilibration of the spout and plunger mechanism pushes that deviation to ±3–6 g per section for the first 15–30 minutes. On a ten-section double-gob line, sections 1 and 2 might be holding ±1.2 g while sections 8 and 9 are still at ±4.5 g. If the operator is reading a line average from the gob weight system, sections 8 and 9 are invisible until blowouts start.

Section-by-section gob weight logging in the first 30 minutes after restart is not optional. It is the data point that tells you which sections have thermally stabilised and which haven't. Without it you are managing line averages that conceal exactly the variance you need to find.

This recovery window extends further in plants where cullet purity is inconsistent. In parts of the GCC and in Egyptian container glass operations, cullet use typically runs 55–70% versus the EU norm of 70–90%. Melt homogeneity fluctuations under those conditions extend the gob temperature recovery window section by section. Post-changeover pack-rate recovery times of 60 minutes per section are common in these environments. A changeover plan that assumes a 45-minute stabilisation window in that context sets the crew up to miss targets before they start.

Mould library gaps are eating your changeover and nobody is counting them

Here is a number that surprises plant managers until they see the data: 18–28% of total IS machine changeover duration is attributable to missing, misidentified, or unserviced tooling. Not to slow forming setup. Not to furnace variability. To tooling management.

Each section change requires at minimum six tooling types: blank, finish, neck ring, plunger, baffle, and bottom plate. On a complex NNPB job, that number reaches ten tooling types per section. On a ten-section line you are coordinating 60 to 100 individual pieces, staged, checked, and serviceable before the line comes down. Plants without serialised mould tracking at the mould prep bench discover missing or unserviced tooling during the changeover, not before it. That discovery does not subtract from setup time. It adds to it.

Saudi Glass Manufacturing in Dammam operates Emhart-Glass IS machine lines without resident IS machine engineers on site. Major job changes rely on OEM technicians flown from Europe, adding an estimated 12–24 hours to planned changeover windows compared to European plants with trained in-house crews, and adding USD 15,000–25,000 per changeover event in contractor and lost-output cost. That differential does not come from the IS machine. It comes from accumulated tooling and knowledge management gaps that a structured mould library and in-house skill programme would close. The equipment is the same. The system around it is not.

Look, the data says one thing and the floor says another, and in most plants the floor wins because the data is not being collected. A serialised tooling library and a pre-change tooling audit completed before line-down converts what is currently an internal disruption into an external task, recovers the 18–28% of changeover time currently lost to tooling searches, and does it with zero capital expenditure. This is the principle behind a structured Job Change Tool: tooling readiness is owned, audited, and signed off in the preparation phase, not discovered when the line is already down. For a common vocabulary across shifts and mould prep benches, the glass glossary gives teams a single reference standard for each tooling component that survives staff turnover.

The carbon cost dimension adds another reason to close this gap quickly. Under EU ETS Phase IV, a 400 t/day furnace running through a 12-hour unmanaged changeover with zero productive output incurs approximately EUR 200–320 in direct EUA carbon cost before you add gas, labour, or lost contribution. That figure is now auditable under Directive 2023/959/EU's raised linear reduction factor of approximately 4.2% per year from 2024. Avoidable changeover losses are becoming carbon liabilities, not just efficiency gaps.

The changeover that runs nine hours on a line capable of five hours is not a furnace problem. It is not an IS machine problem. It is a sequence problem, a preheat problem, a timing re-entry problem, and most of the time a tooling readiness problem. All four are fixable without new capital. A vendor-neutral container glass consultant running a forming audit across your IS machine lines is the practical starting point. What you find in the first two days typically tells you more than 12 months of cold-end defect data. So what is stopping your plant from running that audit this quarter?