Where Container Glass Plant Overtime Cost Actually Hides

-65% overtime. -42% absenteeism. $1.7M AUD back on the P&L inside 12 months. That's what happened at a three-line Brisbane container glass plant I worked in 2014, once we stopped hiring our way out of the problem and started diagnosing it. None of it came from a new furnace or extra headcount. It came from discipline.

Every plant manager I talk to treats overtime as a staffing line. Roster harder, add a crew, sweeten the weekend rate. None of that touches the real driver. In a well-run container glass operation, furnace OEE should sit around 85-92%. Drop below 80% and most plants don't fix the availability loss, they just throw overtime at the tonnage gap. That's expensive cover for a problem nobody has actually named.

Why container glass plant overtime cost isn't a staffing line

Overtime in this industry is almost never a headcount shortage. It's a symptom that shows up hours after a hot-end decision went unmanaged. A forehearth that's drifted 2°C off its zone target doesn't announce itself as a hot-end problem. It shows up six hours later as a spike in blister and stone rejects at the cold end, and someone authorises an unplanned inspection shift to sort it.

I've audited plants where the P&L books that as "cold-end labour variance." It isn't. It's a process control failure wearing a payroll disguise (and no, blaming the cold-end crew for the reject rate doesn't fix the forehearth).

Overtime isn't a labour cost. It's what a plant pays for not standardising the shift before it.

Where the hours actually hide

Four places, almost every time:

• Job and mould changeovers that blow out from a 90-120 minute target to three-plus hours because gob weight and take-out timing weren't pre-staged
• Checkmark and cord defects that force a hold-and-inspect protocol, idling the lehr and packing crews downstream until the batch clears
• Hot-end repairs, an electrode change or a refractory patch, that run past shift handover and trigger 1.5x-2x overtime for the outgoing crew
• Reactive call-ins after forehearth gob temperature drifts outside a ±1°C band, producing intermittent stone and blister defects nobody catches until the cold end does

On an old Emhart eight-section machine running 1980s-era controls, a three-hour changeover barely raises an eyebrow. On a newer Heye Smart H1 line, it should be a five-alarm fire.

And most plants don't even log the gap as changeover variance. It disappears into "planned maintenance" instead, and nobody goes looking for it again.

The Brisbane numbers, and the discipline behind them

The 2014 plant had a 0600 handover that missed night-shift swabbing data on roughly 70% of changeovers. We confirmed it by pulling three months of logs against what operators actually reported verbally. Nobody was lying. Nobody was writing it down either.

Look, the data said one thing and the floor said another, and reconciling those two is most of what this job actually is.

We fixed ownership first. The hot-end superintendent owned recipe lock, full stop. An operator did not touch a set point without sign-off, and that sign-off got logged, not radioed. Gob weight CV had to hold inside 0.4% before a mould change was declared complete, not "close enough for first ware." The changeover checklist stopped living in someone's head and started living in a system every shift could see.

That's the mechanic underneath what's now the Job Change Tool: a versioned SKU library so the recipe and mould spec are locked, not remembered, and a live execution checklist mapped against the nine-stage Job Change Lifecycle so a supervisor sees exactly which stage a line is stuck in, in real time, instead of finding out at the end of a blown shift. Call it SMED for container glass if you like the shorthand. What it actually does is remove the excuse for improvisation.

Middle East plants carry the same disease with a different trigger

GCC container glass plants don't fight the same energy cost pressure European operators do. Industry energy pricing data has Gulf furnace gas running $1.25-3.00/MMBtu against €25-40/MWh gas-equivalent costs in Europe. That advantage doesn't make overtime disappear. It just shifts the pressure sideways, onto labour and changeover discipline, because energy isn't the constraint eating the P&L out there.

Saudi Arabia's Nitaqat programme adds another layer. Under the Saudi Ministry of Human Resources and Social Development's Saudization quotas, plants are staffing a rising share of hot-end and cold-end roles with nationals still climbing the IS machine learning curve. That's not a criticism of the policy. It's a training-ramp reality, and it shows up as coverage overtime while newer operators get to competency on gob weight control and take-out timing.

The USA carries its own version of the same problem. Several operators I've spoken with have watched recorded overtime climb after reclassifying hot-end leads as non-exempt under Department of Labor FLSA threshold updates, with zero change in actual output. The number on the P&L moved. The furnace didn't run any differently.

What generic consultants miss

Most OEM-affiliated advisors benchmark a plant against the machine builder's own rated changeover speed, assuming ideal mould and gob conditions that stopped existing 20 years into the furnace campaign. Their recommendations under-count the real variance drift actually driving the overtime. A generic Lean or Six Sigma boutique tends to model overtime as a rostering problem, missing that the dominant driver is usually a hot-end process variable, cullet ratio, forehearth temperature, plunger cooling air pressure, surfacing as a cold-end labour line hours later.

That's the gap a vendor-neutral container glass consultant is built to close: no equipment upgrade to sell, no rostering software to push, just the plant's own process data and a systemised way to act on it. Our founder, Zaid Hassoneh, started on the floor at O-I Brisbane in 2005 and was Plant Manager there by 2019, before leading the $220M USD Arglass Yamamura greenfield build in the USA from 2019-2022. His Brisbane plant manager experience is where most of this playbook actually got tested, not written. We run a version of the same diagnostic as our management audit, and overtime is usually the first number that moves.

23 months after that Brisbane fix, the plant manager told me overtime hadn't crept back. Not because nobody worked a hard weekend since then. Because the system doesn't depend on who's rostered on anymore. That's the whole point.