Swabbing Programmes That Beat the Cross-Shift Drift

It's 0600 on a two-furnace plant and the night-shift IS operator is walking the line one last time before handover. He's behind on his swab interval, so he goes heavier than usual on cavities four and six before he clocks out. Nobody logs it. Twenty minutes later the cold end starts pulling checks off section six at a rate that makes the QA tech reach for the phone.

That handover moment is where most swabbing programmes actually fail. Not in the formulation. Not in the swab stick or the spray head. In the thirty seconds where one operator's technique hands off to another's, unrecorded, on a chalkboard nobody photographs.

Cross-shift drift is a swab problem before it's a furnace problem

Blank-mould and blow-mould swabbing on an IS machine runs on a fixed rhythm, roughly every 15-30 minutes or every 1,500-2,500 gobs per cavity, using a graphite- or zirconium-based suspension. That part is standard across plants I've audited from Brisbane to the Gulf. What isn't standard is technique: dwell time, film thickness, coverage pattern. Every operator swabs a little differently, and most plants have never measured the difference.

The failure mode has a name: cross-shift drift. The incoming operator, worried about missing an interval, over-corrects with a heavier application at handover. That single heavy swab drops the mould surface temperature at the parison and plunger outside the 500-650°C target band in one step change instead of a gradual trend. The cold end feels it within minutes.

What "on spec" actually means for a swab film

The target film is thin. We're talking a 2-4% suspension in carrier fluid, applied as a few microns of graphite loading, not a visible coating. Over-application past that spec is the leading cause of blister defects, trapped gas under a film that's too thick to release cleanly. Under-application drives stuck ware and accelerates mould wear, because the cavity is running dry between cycles.

This is why the mould-shop technician owns formulation and mixing, and the hot-end operator does not change swab concentration or interval without sign-off. I've seen plants where that line blurs, usually on night shift, usually because someone's chasing a stuck-ware problem with more graphite instead of asking why the interval slipped in the first place. Fix the interval. Don't fix it with more product.

The defects will tell you exactly what went wrong

Checks are hairline crack lines from cold mould contact, your first sign of an under-swabbed or drifted-cold cavity. Blisters point at excess or uneven lubricant trapping gas. Stuck ware means the cavity ran under spec long enough that the parison won't release clean. Washboard marks and cold mould seams both trace back to inconsistent film across the cavity surface, not a formulation problem.

Learn to read the defect before you touch the recipe. A superintendent chasing washboard marks with a mould change is solving the wrong problem if the actual cause is sitting in the swab log, or the absence of one. Our swabbing glossary entry breaks down the mechanism cavity by cavity if you want the full picture.

Track it at the cavity, not the line average

Baseline cold-end reject rate on a well-run line sits around 1.5-2.5%. After an unlogged or off-schedule swab, that spike can hit 3%+ for twenty to forty minutes before it decays back down as the mould re-stabilises, usually over one to two production hours. If you're only reviewing line-average rejects once a shift, that window disappears into the noise. You need swab-log timestamps correlated against cold-end AQL data at the individual-cavity level, not the line level, to catch drift while it's still happening.

In 2021 I audited a two-furnace plant near Dammam where shop-floor temperatures were pushing past 40°C by mid-morning. That heat shortens the mould-to-air cooling differential measurably, which means the swab-interval discipline that works fine on a European or US baseline SOP runs loose in the Gulf. Saudi Arabia's Vision 2030 push has brought serious new capacity online, Obeikan, Saudi Glass, National Glass Industries among others, chasing beverage and pharma demand. Pharma and premium spirits re-export work through Jebel Ali doesn't tolerate a 3% reject spike. The SOP has to account for the climate it's actually running in, not the one it was written for.

A compliant swab log and a stable mould temperature are two different things. Most plants only measure the first one.

Automation doesn't fix a handover problem

Robotic spray arms reduce interval variance, and on a line with an older Emhart machine running 1980s cam timing next to a chalkboard swab log, that's a genuine upgrade. But I've walked plants that installed a Heye Smart H1-generation robotic arm and saw no movement in their cold-end reject KPI six months later. The arm swabs on schedule. The shift-handover protocol underneath it never changed, and neither did the incentive structure that rewards an operator for looking busy at 0600 rather than logging accurately.

And when a defect persists past the normal one-to-two-hour decay window, don't chase it with a heavier re-swab on the whole cavity. Pull the timestamp log first. Nine times out of ten the persistent problem is a single cavity that's been running off-interval for a shift or more, not a formulation issue across the set (and yes, I know your mould-shop tech says the mix is fine, check the log anyway).

Where this fits the bigger changeover picture

Swab discipline is a steady-state problem, but it shows up hardest at the exact moments a plant is already stretched, mould changes, recipe loads, first-ware stabilisation. That's the same window our Job Change Tool tracks section-level variance across, because a plant that can't hold swab technique cross-shift usually can't hold changeover timing either. It's the same discipline gap wearing two different hats.

A vendor-neutral container glass consultant isn't going to sell you a robotic arm and call the job done. The fix starts with logging swab timestamps against cavity-level reject data, closing the gap between what the SOP says and what actually happens at 0600. If you want a second set of eyes on where your programme is actually leaking glass, our hot end audit is built to find exactly that gap.