Vendor-Neutral Container Glass Consulting Under ETS Phase IV

The EU ETS Annual Linear Reduction Factor doubled on 1 January 2024. It moved from 2.2% to 4.3% per year under Directive 2023/959/EU, and most European container glass plants I speak with still haven't stress-tested what that trajectory means for their free-allocation position by 2027 or 2028. Not conceptually. Furnace by furnace, on an actual spreadsheet, with real carbon cost numbers attached.

That gap is exactly where the wrong adviser costs you money.

The Phase IV numbers look different at furnace level

The EU ETS product benchmark for container glass under Implementing Regulation 2021/447 sits at approximately 0.484 tCO2 per tonne of glass. Plants melting above that carbon intensity receive no surplus allowances and must buy European Union Allowances at market. EUAs averaged roughly €62 per tonne in Q1 2025. On a 300 tonne/day air-fuel furnace operating above the benchmark, that unmitigated exposure runs €2.5–3.5 million per annum, and it compounds each year as the reduction factor cuts free-allocation volumes through to 2030.

CBAM adds a second dimension. Part of the EU Fit for 55 package, it requires importers of container glass under CN codes 7010.10–7010.90 to purchase CBAM certificates at the prevailing EUA auction price from 1 January 2026. That closes the carbon cost arbitrage window Turkish and North African producers have relied on. For European plants it is partial relief on competition. It doesn't solve the abatement question on your own furnaces.

The EU Glass Manufacturing BREF, revised in 2023, sets a BAT-associated energy consumption level of 3.3–4.6 GJ per tonne for container glass furnaces. Plants consistently exceeding 4.6 GJ/tonne face enforceable improvement conditions under Industrial Emissions Directive Article 14 permit reviews. European furnaces running oxy-fuel with cullet above 70% are already at 3.2–3.6 GJ/tonne. The industry average on air-fuel plant with mixed cullet sits around 4.2 GJ/tonne. A furnace above 4.8 GJ/tonne is a primary capital trigger under Phase IV economics. Knowing where your plant sits, with an accurate carbon cost model attached, is the starting point for any credible investment case.

Generic advisers don't know what a seed defect is

In 2021 I was reviewing an energy management report commissioned by a three-furnace container glass plant in northern Spain. The report recommended a straight cullet ratio increase from 55% to 75% as the primary ETS compliance lever. No mention of batch redox adjustment. No acknowledgement that the plant's forehearth temperature homogeneity was already marginal, running ±4–5°C across the orifice ring cross-section when the target is ±2°C.

They implemented the increase. Within six weeks the plant was seeing elevated seed counts on two lines. Seeds are gaseous micro-inclusions under 1 mm from incomplete sulphate fining. A 10°C drop in fining zone temperature, from 1,480°C to 1,470°C, routinely increases seed counts by 15–30% within a single eight-hour shift. The energy consultancy attributed the defect spike to batch moisture. Not a fining chemistry problem, apparently. It was the interaction between cullet chemistry, sulphate fining behaviour, and a forehearth nobody had profiled in two years.

And it's not just energy consultants. OEM-affiliated advisory practices structure assessments around capital procurement from their parent company's portfolio. IS machine upgrades. Combustion systems. Inspection hardware. The question of whether operational tuning (forehearth profiling, gob weight optimisation, IS timing calibration) could deliver equivalent pack-to-melt improvements at less than 10% of the capital outlay rarely makes the final recommendation. That conflict of interest is structural, not incidental, and it is rarely disclosed in the engagement letter.

Cullet is an ETS lever, but only if you manage the chemistry

Each 10% increase in cullet substitution typically reduces specific energy consumption by 0.25–0.35 GJ per tonne and cuts Scope 1 CO2 intensity by approximately 11–15 kg CO2 per tonne. FEVE reported EU-27 container glass production at approximately 21 million tonnes in 2023, with average recycling rates exceeding 80.8% across the EU-27. Germany, Belgium and the Netherlands each exceeded 90%, improving external cullet quality available for closed-loop remelting. That infrastructure exists in northern Europe in a way that doesn't transfer to most other markets.

But cullet quality management is where this gets complicated fast. Metallic contamination above 3 mm, or ceramic fragments in external cullet (and yes, your cullet supplier's certificate may well say otherwise, so run the spectro analysis before you adjust the batch sheet), are the primary driver of downstream stone defects. Stones are crystalline inclusions from refractory dissolution or unfused batch carry-over. A single AZS crown stone event on a 12-section IS machine running at 200 bpm can trigger a 4–6 hour section shutdown, directly cutting your daily pack-to-melt ratio by 1–2 percentage points.

Every 1% pack-to-melt gain on a 300 tonne/day furnace represents approximately 3 tonnes of glass saved per day. That's roughly 15–16 GJ of avoided melting energy, worth €250–350 per day at current European gas prices, before you count the ETS benefit. The cullet decision and the defect management decision are the same decision. You can't run them as separate workstreams and expect the maths to hold.

The carbon model and the production model have to be the same model. A carbon consultant who has never had to explain a stone defect to a customer service team will get the abatement projection wrong.

What a vendor-neutral engagement actually delivers

Verallia reported a Scope 1+2 CO2 intensity of 204 kg CO2 per tonne of glass packed in FY2023 and targets 150 kg per tonne by 2030. Reaching that trajectory requires electrification or hydrogen co-firing beyond what operational optimisation alone can deliver. Operational optimisation is what builds the credibility of the investment case, though. You don't go to a board asking for electric boost capex on a furnace running at 4.5 GJ/tonne with a 1.8% reject rate and no explanation for the variance. You go after you've done the floor work.

A vendor-neutral container glass consultant doesn't arrive with a preferred combustion supplier or an OEM relationship that needs protecting. On the hot end, the assessment means forehearth profiling, pack-to-melt tracking against a real baseline, and IS plunger timing audits. On an older Emhart Series machine running pneumatic timing controls, drift above 3 ms in the stroke window can go a full shift without anyone logging it. On a modern Heye International servo-electric line, the alarm fires within minutes. Either way, the target stroke window on an NNPB process is ±1.5 ms, and when it drifts you get thin-wall sections and hot-end vision system rejects that trace back to timing, not forming.

On the job change side, cross-shift variance on identical SKUs still runs 30–60% in European container glass plants that haven't systemised the changeover. The Job Change Tool we use at Lean Glass maps execution across a 9-stage lifecycle with section-by-section progress visible in real time. A hot-end superintendent who owns recipe lock and documents every set-point deviation across shifts builds the data set that separates a real abatement baseline from a reported one. Most plants I audit are relying on the reported number. The real number is routinely 4–8 OEE points lower.

O-I Glass announced a hydrogen co-firing trial at a German plant in 2024 as part of the HyGlass consortium. Ardagh flagged constraints on European decarbonisation capex with net debt/EBITDA above 5x. Both signals point in the same direction: the margin for error on Phase IV investment decisions is tight. It's not glamorous work, building the operational baseline that supports the capital case. It's the work.

If your Phase IV compliance review is coming up and you're not confident the abatement model reflects what's achievable at plant level, start with a structured operational assessment. Our hot end audit programme gives you that operational baseline in 5–10 days on site. Our strategic advisory work connects it to the investment case your board needs to see. Both are described on our vendor-neutral consultancy page.