What pack-to-melt actually measures
Pack-to-melt ratio = (saleable ware packaged in tonnes) ÷ (total glass melted in tonnes). It is the most honest efficiency KPI in container glass because it captures every loss between molten glass and saleable pallet. Forming rejects, takeout breakage, conveyor breakage, lehr losses, cold-end inspection rejects and packaging losses all show up in this one number.
Why it beats OEE for container glass
OEE is a useful rollup but it has structural blind spots. Performance × Availability × Quality each rely on careful definitions; the same OEE number can hide very different operational realities. Pack-to-melt is harder to flatter — the cullet trail does not lie.
How to measure it correctly
- Numerator: saleable packed ware in tonnes (not units)
- Denominator: total glass melted in tonnes (batch + cullet × campaign efficiency)
- Time period: minimum a full month, ideally a full campaign, to wash out daily noise
- Boundary: through-the-fence — saleable pallet leaving the plant counts; held inventory does not
Common ways plants flatter the number
- Counting hot-end output as numerator (excludes lehr and cold-end losses)
- Counting batch as denominator without cullet input
- Excluding 'planned' rejects (job change first ware, sample, etc.)
- Reporting weekly, where short-term variance washes out the structural number
What good looks like
- Best-in-class beverage glass plants: 88–92%
- Mid-range plants: 80–86%
- Plants with persistent issues: 70–78%
- The variance is bigger than most boards realise
How to lift pack-to-melt
The biggest pack-to-melt lifts come from job change discipline, hot-end audit findings, and ware handling improvements — in that order. A 1-point pack-to-melt lift at a medium plant ($150M revenue, 15% margin) is typically worth $3–8M annualised EBITDA. Many plants leave 4+ points on the table at any given moment.