Why the hot end is where OEE lives or dies
In a container glass plant, the hot end is the narrow waist of the hourglass. Everything upstream — batch house, melter, forehearth — delivers to it. Everything downstream — annealing, cold-end inspection, packaging — depends on what comes out of it. Optimise the hot end and the whole line runs better; leave it alone and nothing else you do really matters.
This is the checklist we actually run when we walk onto a forming floor for the first time. It is not exhaustive — no checklist is — but it covers the 80% that explains most of the variance between plants that run well and plants that firefight.
1. Forming: the fundamentals, honestly assessed
- Section speed balance. Pull speed uniformity across sections tells you whether your forming is constrained by the slowest section or whether an average hides a problem. Film a full job and tag any section deviating more than 3% from the target.
- Defect pareto by section. Not by line, not by shift — by section. Eight of ten plants we visit do not have this report, or they have it but nobody looks at it daily.
- Swab cycle discipline. Ask the operator to show you the current swab cycle and the record of the last ten cycles. If they cannot show it in under 60 seconds, you have a problem long before you have a defect.
- Reject chute utilisation. How often is the chute active? What defect triggered it? Is the operator tagging root cause or just clearing it?
2. Forehearth: the boring system that decides quality
Forehearth is rarely glamorous, rarely cited in boardrooms, and almost always where we find more opportunity than anywhere else on the hot end. The questions:
- What is the delta between setpoint and actual temperature at the gob delivery point over the last 24 hours? If it's ±2°C, you have steady glass. If it swings 10°C, your forming is working around a moving target.
- Is the cooling air distribution balanced across the gob? Gob shape stability tells you instantly.
- How often is the forehearth tuning reviewed for each product? If it's once a year, you are optimising an imaginary plant.
3. Line coordination: the conversation between crews
The best forming and forehearth in the world still lose ground if the hot end, cold end, and packaging are not coordinated. The diagnostic is observational:
- Watch a crew handover. How long does it take, and what is transferred — just numbers, or context and decisions?
- When a jam-up happens in packaging, how fast does the hot end know? How fast does it change behaviour in response?
- Is there a single operating rhythm — a tier-one huddle every 2 hours, say — or does every area run on its own tempo?
4. Data: does the plant know what it is doing?
The number we ask for first is not OEE. It's the answer to: "Between hour 3 and hour 4 of the last job run, what was the defect rate on Line 2, Section 5, by defect mode?" If that takes under five minutes to answer from live data, the plant has good digital bones. If it takes days, no amount of coaching will hold an improvement in place.
5. Changeovers: the recurring reset
We covered this in depth in our post on sub-six-hour job changes, but the quick diagnostic: what was the changeover time and first-hour yield on the last five job changes? If there is no consolidated record of both, that is itself a finding.
The two-hour assessment
You can do a meaningful first pass on the above in two hours of focused floor walking plus one hour with the data team. If we are invited in, we spend 3–7 days doing this properly — covering every section, every shift, and at least two full job changes. But the two-hour version is enough to tell a leadership team whether they have a genuine problem or a manageable one.
What happens next
The output of a hot end diagnostic should be a prioritised action plan — owner, deadline, KPI per item — not a bound report. If the first action is not on the floor within a week of finishing the diagnostic, the engagement is already losing momentum. We will write more on the diagnostic-to-action handoff shortly.