The Four Forming Defects To Catch In The First Ten Cycles

Twenty-three minutes. That's how long it took a section on a five-line plant I audited in Georgia to go from clean ware to a 6% check-line reject rate, and nobody caught it until the cold-end inspection data got pulled at end of shift. Not a furnace problem. A mould-temperature problem, sitting in plain sight in the first ten cycles, that nobody was required to log.

That's the whole argument for a disciplined first-ten-cycles check. After any job change, mould swap, or stop over fifteen minutes, you get a narrow window where the forming process tells you exactly what's wrong before it becomes a packing-hall problem. Miss that window and you're chasing the same defect three days later as a mystery.

Why the first ten cycles matter more than the next ten thousand

Standard practice is a 100% visual and dimensional check on the first ten cycles post-changeover. That means every piece gets checked, not sampled, for wall-thickness distribution, verticality, and finish sealing-surface ovality against blueprint tolerance, typically ±0.13mm on the critical finish dimensions. Nobody enjoys doing this by hand at 2am, but it's the cheapest quality insurance you'll ever buy.

The reason it works is that job-change defects cluster at four dominant failure modes, and each one has a distinct signature you can read fast if you know where to look. I've run this checklist on Emhart NIS lines with 1980s cam boxes and on a recent Heye Smart H1 install, and the physics doesn't change even though the controls do.

Cold mould checks tell you the mould wasn't ready

Cold mould checks show up as radial fracture lines running out from the finish, and they're a thermal-shock signature, plain and simple. The mould went into production before it reached temperature. Blank mould operating band typically runs 480-560°C depending on container weight and glass composition, and if you've got more than 15-20°C variance section to section, you will see check-line clusters concentrated on the cold sections.

I worked a two-furnace plant in 2014 where the night shift was releasing jobs off a fixed timer instead of an actual temperature read. Every job change under-preheated section 4 by roughly 30°C because the swab cycle ran short on that one station. The fix cost nothing. It was a five-minute infrared gun check added to the changeover checklist, countersigned by the shift quality technician before release.

Stuck ware and blisters point in opposite directions

Stuck ware tears the glass surface as it separates from the mould, and it's usually swab interval or mould temperature below roughly 450-480°C at the blow side. Blisters and reboil look completely different under a polariscope, trapped gas bubbles near the parison wall, and they trace back to refractory condition or batch moisture, not mould temperature at all. Operators who don't separate these two get sent chasing the wrong root cause for a shift.

A checklist tells you what to look at. It doesn't tell you what you're looking at. That's still the operator's job.

Gob weight consistency matters here too. Variance beyond ±0.5g from target is a common trigger for both stuck ware and thin-wall rejects, because an underweight gob cools faster in the mould and behaves like a cold-mould problem even when the mould temperature reads fine. Check gob weight before you start chasing mould thermocouples.

Choke and thin-wall faults are a timing problem, not a glass problem

Uneven gob temperature or misaligned plunger timing produces choke and thin-wall distribution faults, and they read on a wall-thickness gauge as an asymmetric profile rather than a clean radial pattern. This is where IS machine section timing earns its place on the checklist. Cam and valve timing drifting outside ±2° from master setting is a leading root cause of both check-line and stuck-ware defects, and a section running out of tolerance should stop and requalify, not run to packing on a hope.

• Cold mould checks: radial fracture from the finish, thermal-shock signature
• Stuck ware: torn surface at separation, swab or temperature root cause
• Blisters/reboil: trapped bubbles near parison wall, refractory or batch moisture
• Choke/thin-wall: asymmetric wall profile, gob temperature or plunger timing

And this is the part most outside consultants skip. A mis-tuned automatic inspection threshold on the cold end, an Iris or iQC system set too loose, can mask a hot-end problem for hundreds of cycles before anyone sees it. I've seen plants treat hot-end changeover and cold-end inspection settings as two different departments' business. On the floor they're the same problem viewed from two ends of the lehr.

The handover gap that undoes all of it

None of this matters if the checklist doesn't get signed. Shift-level accountability should sit with the hot-end operator to execute and log the first-ten-cycle check, with the line supervisor or shift quality technician required to countersign before the job releases to full rate. On the plants I audit, this sign-off step is the one most commonly skipped, and it's the one that OEM-affiliated service reports rarely even mention as a control point, because it's free to fix and there's no equipment sale attached to fixing it.

The 0600 handover misses the night-shift swabbing data on a surprising number of lines I see, easily seven times out of ten. Don't forget the lehr either. Container weight and wall-thickness changes at job change shift the required residence time, and if annealing entry temperature and belt speed aren't re-verified before the first ten cycles reach the lehr, you get delayed-fracture breakage discovered on a pallet three days later, long after anyone remembers the job change happened.

What this is worth

Typical hot-end reject rate at steady state runs under 0.3%. Unqualified changeovers routinely spike to 3-8% in the first cycles. That gap is where the money sits, and it compounds. Every rejected container is remelted energy and re-embodied emissions, a cost that regulatory-compliance advisers focused purely on EU ETS Phase IV or CBAM exposure rarely connect back to the forming floor, even though the Linear Reduction Factor cutting free carbon allowances by 4.3% annually through 2027 makes every remelt more expensive to carry on the books.

This is exactly the territory the Job Change Tool was built to systemise, mapping first-ware checks and sign-off points to the 9-stage Job Change Lifecycle so they don't depend on which supervisor is on shift. It's not a generic Lean rebadge with a fishbone diagram bolted on. It's built from defect nomenclature operators actually use, checked against a forming audit process that's been run on real lines, not written in a conference room.

Run your own first-ten-cycles data for a month before you decide whether this is a training problem or a systems problem. My money's on systems, most of the time it is.