What Labour Cost Is Doing to US Container Glass Changeovers

It's 0215 on a Tuesday and a two-furnace, six-section plant in the US Southeast is 40 minutes into a job change. The hot-end superintendent has gone home sick. The operator running the change has done this SKU once, eight months ago, and the last forming specs live in a notebook that is currently in the absent superintendent's truck.

That notebook is a single point of failure worth roughly $180,000 in downtime per shift. Nobody in that plant called it a labour cost problem.

The wage bill is the visible cost, not the real one

When plant managers across US container glass operations talk about labour cost pressure, they mean the wage line. Hourly rates for skilled IS machine operators have climbed hard across the Southeast and Midwest since 2020, and retention bonuses that barely existed five years ago are now standard at most major producers. That's real pressure and it belongs in the budget conversation.

But the wage line is the cost you can see. The cost you can't see is what happens when you run a job change with the wrong person, the wrong recipe, or no documented recipe at all. Cross-shift variance on identical SKUs in plants without a systemised approach typically runs 30-60%. That's not a furnace problem or a maintenance problem. It's a knowledge problem, priced out per changeover event, every shift, every week.

A 1% OEE improvement on a standard five-section container glass line in North America is worth several million dollars in EBITDA depending on mix and throughput. Labour cost pressure makes that number matter more, not less.

The skill drain is accelerating and the maths are getting worse

The US container glass market has gone through significant capacity rationalisation over the past several years. That rationalisation hasn't reduced the complexity of job changes at surviving plants. In many cases it has increased it, as plant managers absorb more SKUs onto fewer lines with the same or smaller crews.

A meaningful cohort of experienced hot-end operators who joined during the capacity build of the 1990s will have reached retirement age by the end of this decade. What they carry out the door is rarely documented anywhere useful.

In 2018 I was brought into a two-furnace plant in the US Southeast to look at their changeover performance. The plant ran four SKUs on a single line, and three of those four job change procedures existed only as memory inside one operator's head. He'd been there 22 years. When I asked the plant manager what the plan was for that knowledge, he told me they'd been meaning to write it down.

Not documented. Meaning to. Look, no plant manager I've met thinks tribal knowledge is a good system. They just haven't had the time to fix it, and then one day the operator retires and the time has passed.

The plants that feel the skill drain least are the ones that locked their forming specs before the experienced operators left. Gob weight CV targets written down at ≤ 0.4%. Mould preheat curves documented at 480°C ±10°C. Section timing tolerances held to within ±10ms and version-controlled, not living in somebody's personal notes. The plants that skipped that work are training newer operators against knowledge passed sideways across shifts, and the first-ware quality reflects it within the first production run.

A lot of these plants are running perfectly serviceable Emhart B9 machines with updated inspection systems and modern section controls. The equipment isn't the constraint. The knowledge around the equipment is.

Every container glass plant has a job change procedure. In most of them it lives inside one person's head, and that person is thinking about retiring.

Changeovers are where the gap shows up first

Baffle marks are one of the early tells on a plant with a training problem. An operator who hasn't been shown the proper centring tool procedure will set the baffle by feel, and you'll see the marks within the first 30 minutes of production ware (and yes, I know your maintenance team will call it a temperature issue first. Check the centring tool reading before you touch the forehearth profile).

The 0600 handover is another diagnostic. On most US lines I've audited, the outgoing night shift misses the swabbing frequency data at least 70% of the time. That looks like an admin failure. It's actually a training failure dressed as a process failure. When the outgoing operator hasn't been trained on what to hand over, they hand over nothing useful, and the day-shift superintendent starts their run blind.

The hot-end superintendent owns recipe lock in a well-run plant. The operator does not change set points without sign-off. That structure holds when the superintendent has been in the role long enough to understand the SKU library. When superintendent tenure runs 18 months because you can't retain them on current wages, that ownership collapses. New superintendents inherit chaos, and the operators fill the gap by making their own calls on the floor. The downstream cost of that is settle waves and choked necks on the first ware, which adds 20-30 minutes to time-to-stable-pack on a changeover that was already running long.

You cannot train your way out of an undocumented process

This is the part most US plants get wrong. The natural response to skill drain is more training: more inductions, more classroom hours, more OEM certification programmes. Then the new operator gets on the floor and the procedure in the training material doesn't match the job aid on the IS machine, which doesn't match what the senior night-shift operator actually does.

And the changeover runs 95 minutes instead of 45. Every time. For every new operator. Predictably, not randomly.

The answer isn't more training hours. The answer is building the process first, then training into it. That means locking every SKU recipe, every mould set specification, and every section timing target into a single versioned source before putting a new operator in front of the machine. It means a live execution layer that walks the operator through the 9-stage Job Change Lifecycle in sequence, with section-by-section progress visible to the superintendent in real time. It means KPI tracking that surfaces cross-shift variance by operator and by SKU, so you can see exactly where the skill gaps are rather than guessing.

That's what the Job Change Tool is built to do. Not a training platform. A systemised process that makes knowledge transferable across shifts, across operators, and across the turnover cycles that US container glass plants are now living with. It's vendor-neutral, not tied to any OEM, built by operators who ran the lines. The cross-shift variance data surfaced over 90 days at a typical plant is usually enough to identify two or three sections where targeted retraining will yield measurable improvement. When we work with US plant managers as an independent container glass consultant, the finding is the same in plant after plant: knowledge that was never written down, running inside the heads of people who are about to leave.

A management audit will show you exactly where your exposure sits before the next experienced operator hands in their notice.

23 minutes. That was the average reduction per changeover event we tracked after the process was documented and locked at one four-section US plant. Across shifts and SKUs, that number changes the EBITDA conversation entirely.

Which brings the question back to your plant: is the knowledge your experienced operators carry written down somewhere it will survive their departure, or is it still sitting in someone's truck?