Mould Rotation Calendars Across A Multi-Plant Network

Ask a plant manager who owns the mould rotation calendar and most of them point down the corridor at the mould shop. That's mistake number one, and it's the mistake that quietly eats capacity across a multi-plant network long before anyone notices the OEE numbers slipping.

A container glass mould rotation calendar isn't a maintenance list. It's an asset allocation decision, and asset allocation decisions belong on the plant manager's desk, not buried in a spreadsheet the fitter updates when he remembers.

What mould inventory drift actually costs you

Here's the pattern I see on nearly every multi-plant audit. Plant A has three sets of a fast-moving 750ml mould rated for a Heye Smart H1 line. Plant B, 200 kilometres away, is running the same SKU on an older Emhart IS machine still on 1980s cam timing, and nobody has reconciled which sets belong where. Somewhere in the middle, a mould set with 1.1 million cycles on it gets scheduled for another run because the tracking sheet says "available" instead of "due for recut."

That's not a hypothetical. In 2018 I audited a four-furnace GCC operation running a mixed OEM fleet, three different machine builders across five lines, and the mould rotation calendar was three separate spreadsheets that didn't talk to each other. The mould shop foreman was making recut decisions on gut feel (and to be fair, he'd been doing it that way for fifteen years and nobody had told him different) because nobody had given him a cycle-count trigger.

We pulled the cycle logs and found two mould sets running past 900,000 cycles with zero recut flag. Result: baffle marks across most of a shift's run, and a first-ware quality drop the plant blamed on the batch house for three weeks before anyone thought to check mould condition.

That's the real cost of drift. Not a catastrophic failure. A slow bleed of first-ware quality and unplanned micro-stops that never shows up as a single line item, because it's spread across every section on every line running that SKU.

A mould rotation calendar that lives in one plant's head office is not a network asset. It's a liability wearing a spreadsheet.

The specs a rotation calendar has to lock down

A rotation calendar earns its keep when it tracks four things at the individual mould-set level, not the SKU level: cumulative cycle count against a recut trigger (typically 800,000–1,000,000 cycles depending on glass colour and cavitation), preheat curve compliance at changeover (target 480°C ±10°C before first ware), swab interval against actual run hours, and which plant and line the set is physically sitting on right now.

Miss the third one and you get swab burn on one line while the identical set two plants over is under-swabbed and throwing checked necks. I've seen both problems on the same mould family in the same quarter, at two plants owned by the same group, because nobody was comparing notes. Get this wrong and it's not just a mould problem, it's a changeover problem too, since every mould swap sits inside the mould-change stage of a plant's job change sequence, and a set that turns up short on preheat compliance turns a clean changeover into a two-hour scramble.

The hot-end superintendent should own the recut trigger decision. The mould shop owns physical condition and preheat compliance. The plant manager owns which set goes where, because that's a capacity call, not a maintenance call. When those three lines blur, and they usually do, the rotation calendar stops being a calendar and turns into an argument.

Multi-plant networks make drift worse, not better

Single-plant operators can get away with tribal knowledge for a while, right up until the shift that knows where everything is retires. Multi-plant networks don't get that grace period. A mould set travelling between two sites crosses a paperwork gap almost every time, and the 0600 handover is where it usually falls through. On more than half the changeovers I've reviewed, the incoming shift has no record of the outgoing shift's swab count or the last recorded mould temperature reading.

And when a set moves plant to plant for a shared SKU, the receiving line often runs a different forehearth profile and different job change habits. One plant preheats at 470°C because "that's what works here." The other runs 490°C. Neither number is wrong on its own. Together, across a shared mould set, they're a stones and cords problem waiting to happen.

Not a furnace problem. A tracking problem.

Why this belongs with asset positioning, not the mould shop

This is exactly why we treat mould rotation as an asset positioning question, not a maintenance schedule. Which mould sets sit at which plant, on which line, running which SKU, is a capacity decision with real throughput riding on it across a multi-plant network. Get the positioning wrong and you're paying for idle capital sitting in a warehouse while a line three sites away runs a set past its recut trigger.

It's not glamorous work. It's the work. A vendor-neutral container glass consultant looking at a network for the first time will usually find the rotation calendar before they find anything wrong with the furnace, because the calendar is where the tribal knowledge concentrates. Fixing it is often a strategic advisory conversation before it's ever a spreadsheet fix, because it forces the group to agree on who owns capacity decisions across sites.

Zaid Hassoneh saw this exact failure mode from the inside, running plants for O-I before leading the $220M USD Arglass Yamamura greenfield build in the USA. A single-plant mould tracking habit doesn't scale to a network without someone forcing the reconciliation.

So here's the question worth asking this quarter: if you pulled the cycle count on every mould set running across your network right now, would the number match what's on the calendar? Most plant managers I ask can't answer that inside a week. That gap is your capacity leak.