Container glass capex defence: find the fix before the rebuild

The stone count is the symptom, not the case

Every capex request that lands on a plant manager's desk starts the same way: a stone count that's crept up, a section running hot, a furnace that "feels" old. The rebuild deck goes to the board before anyone has separated what the furnace is actually doing from what the batch house and the shift roster are doing to it. That's backwards, and it's an expensive habit.

Stones are the classic capex trigger. A refractory or devitrification particle turns up in the gob, the hot-end team flags it, and within a fortnight someone's costing a cold repair. But a rising stone count is frequently traceable to a specific checker zone or a throat refractory patch, not a furnace-wide failure. Cord and seeds get mis-diagnosed the same way. Both read as "furnace age" on a defect report when the real cause sits in batch house homogeneity or a redox-control drift nobody's trended in months.

In 2019 I walked a wine-bottle plant outside Valencia that had a rebuild proposal sitting on the CFO's desk. The stone count had doubled over six weeks. The furnace was 11 years into a campaign rated for 14. Everyone assumed the refractory was going. It wasn't. A batch-weighing error on the cullet feed, uncorrected for three shifts running, had shifted the redox balance enough to start seeding at the throat. Three weeks of throughput data told a different story than the rebuild deck. Not a furnace problem. A batch house problem. The fix cost a recalibration and a training session, not $20M.

What the free allocation squeeze is actually forcing operators to check first

European producers now have to build two cost curves into every capex case, not one. Under EU ETS Phase IV, the Linear Reduction Factor cuts the overall emissions cap by 4.3% a year from 2024 to 2027, rising to 4.4% from 2028, which tightens the free-allocation benchmark for glass installations every year (European Commission, DG CLIMA). Container glass still sits outside CBAM's current scope, which covers iron and steel, cement, aluminium, fertilisers, hydrogen and electricity, though FEVE has been lobbying for its inclusion on carbon-leakage grounds. That pressure is coming, and plants that wait for it to arrive before addressing furnace efficiency will be negotiating from a worse position.

Here's the lever most plants leave on the table: cullet. FEVE's 2023 figures put European container glass recycling at roughly 80% of packaging placed on the market, with average furnace cullet ratios around 52% industry-wide, and above 90% at the best-run plants. A 10-point increase in cullet ratio typically cuts specific melting energy by 2-3%. A well-optimised container furnace runs 4.5-5.5 GJ/tonne of melt. An under-maintained or ageing one runs 6-7+ GJ/tonne. That swing is large enough to self-fund most operating-fix programmes out of the energy line alone, before you've touched the capex budget (and before you've had the conversation with the board about whether the furnace needs to come down at all).

OEE, changeover, and the points nobody totals up

Industry-wide, container-glass forming lines run OEE around 65-75%. Top-quartile plants run 85%+. That gap doesn't live in the furnace. It lives in changeover time, minor stops and quality losses, which are shift-management and calibration problems dressed up as equipment problems.

Job and mould changeover on an IS machine should run under 20-30 minutes a section. When it drifts past 45-60 minutes, that's not a tooling failure, it's shift-discipline decay, and it shows up as a leading indicator well before it hits furnace-level KPIs. This is exactly the gap our Job Change Tool is built to close: a systemised sequence through the nine-stage Job Change Lifecycle instead of whatever the outgoing shift remembers to hand over.

And the handover is where it actually breaks. On most lines I've audited, the 0600 handover misses the night shift's swabbing data more than 70% of the time. Nobody signs it off. Nobody chases it. It just doesn't move to the next shift, and by the time it shows up as a baffle-mark cluster three days later, the trail's gone cold.

None of this needs new equipment. An older Emhart line running relay-logic job settings can hit the same changeover discipline as a newer Heye machine if the shift structure enforces it. The hot-end superintendent owns recipe lock. The operator doesn't touch a set point without sign-off. That's not a system upgrade. That's a rule, enforced.

• Pull the checker-zone stone report before you cost a reline, not after
• Check achievable cullet ratio against actual, not against last year's number
• Trend changeover variance by shift, not by line average
• Audit the redox curve for drift before blaming refractory
• Compare actual yield to the 90%+ benchmark before comparing furnace age to campaign limit

A furnace doesn't know how old it is. Your shift roster does.

The board doesn't want a defence memo, it wants a number

A cold repair on a mid-size container furnace runs $15-40M depending on tonnage and rebuild scope. That's the number that forces the decision, and it's also the number that makes plant managers reach for a defence memo instead of a diagnosis. Wrong instinct. The board doesn't want a memo explaining why capex should wait. It wants a yield number, an energy number, and a date.

O-I Glass's Fit to Win programme is the clearest public example of this sequencing done at scale, targeting more than $600M in cumulative EBITDA improvement through 2025 via operational and footprint changes ahead of new capital commitments (per O-I Glass investor disclosures), a sequence that included furnace and site closures in Auburn, New York in 2023 and Portland, Oregon in 2024. Ardagh's North American restructuring, including furnace-line idling at its Indiana sites, followed the same order of operations: fix the operating picture, then decide what capital actually needs to move.

Total yield, packed ware over gross pulled, above 90% is considered top-quartile. Hot-end reject and redraw rates around 3-5% are the benchmark for a well-run line. The gap between a plant's actual yield and that benchmark is the single most persuasive argument against premature capex, because it's the one number a CFO can put next to the rebuild quote and ask a real question.

Different regions, same discipline gap

In the Gulf, gas pricing reform has made specific energy a direct margin lever rather than a background metric. Saudi Arabia's tiered industrial gas pricing, adjusted through 2023, has pushed producers like Obeikan Glass and Vetro Middle East to treat furnace efficiency as a line item the CFO watches monthly. Egypt's IMF-linked subsidy reform pushed industrial natural gas toward $4.5-5.5/MMBtu by 2024, squeezing furnace fuel margins at plants including Sphinx Glass and El-Nasr Glass. None of the GCC states run an operating carbon-pricing scheme yet, but Saudi Arabia's 2060 and the UAE's 2050 net-zero pledges are already being used internally by multinational glass groups to justify pre-emptive electrification capex requests that haven't been tested against the operating fix first.

In the US, the sequence O-I and Ardagh have both followed publicly is the one every plant should run privately before a capex paper goes to the board. Not because the furnace doesn't eventually need the money. Because most plants haven't proven they've spent the operating budget's worth of discipline first.

This is the gap a vendor-neutral container glass consultant is built to find, because an OEM-affiliated shop makes its margin on the reline, not on telling you the checker zone was the whole problem. Lean Glass runs on operators who've carried the number themselves. Founder Zaid Hassoneh started on the floor at O-I Brisbane in 2005, ran plants through to 2019, and led the $220M USD Arglass Yamamura greenfield build in the US through 2022, on the back of an O-I Most Improved Job Change Plant Globally award in 2016. That's not a sales pitch. It's the reason the diagnosis starts at the checker zone and not the rebuild quote.

If your capex request is sitting on a desk right now, get the operating case checked before the rebuild case gets written. Our hot end audit is built for exactly that decision point, and if the number really is a furnace problem, our strategic advisory work will help you take the evidence to the board either way.