The Mould Preheat Curve That Protects First-Ware Quality

It's 0520 on a Tuesday morning. The first gobs are dropping after a job change, section temperatures aren't settled, and the moulds came off the preheat oven twenty minutes early. The operator on section three is calling leaners. The hot-end superintendent is watching the pack chart at the lehr and it hasn't moved.

Nobody in that room is surprised. They've seen it before.

Under-preheating is the most accepted form of first-ware waste in the plant

Most plants treat mould preheat as a box to tick, not a process to control. The oven goes on at T-3 hours, someone checks that the moulds feel warm enough by palm, and the job change proceeds. On paper, preheat happened. On the floor, what happened was an uncontrolled ramp to an unknown temperature followed by a job change that will bleed rejects for the first 45 minutes of production.

The target for any sound container glass mould preheat strategy sits at 480°C ±10°C at mould contact surface before the first gob drops. That number isn't arbitrary. It's the threshold where the thermal shock gradient between a 1100°C gob and the blank mould stays within the range the glass can absorb without the surface tension differentials that cause settle waves and chilled surface defects. Drop that contact surface temperature to 400°C and you'll see it in your stones count and cold checks within the first 200 pieces.

I've watched this play out on three continents. The failure mode is always the same: someone pulled the moulds off the preheat oven early because the machine needed to run, and nobody wrote it down as the root cause.

What a controlled preheat curve actually looks like

The ramp rate matters as much as the final temperature. Taking moulds from ambient to 480°C in two hours puts thermal gradient stress through the mould body itself. Cast iron absorbs heat unevenly, and if you've got thicker baffle sections in the mould set, those will lag by 20-30°C compared to the blank body during a fast ramp. That asymmetry shows up as baffle marks on first ware, and operators misread it as a baffle alignment problem. They adjust centring, the marks persist, and the real cause sits untouched (this is one of the more reliable ways to waste an hour of gubbins time on a fresh job change, and I've done it myself, which is how I know).

A controlled ramp of 80-100°C per hour to a 420°C hold, followed by a slower 40°C per hour final climb to 480°C, gives the mould body time to equalise across the section. Total ramp time is around four hours. Soak at 480°C ±10°C for a minimum of 45 minutes before the first mould goes on the machine.

In 2017, I was part of a changeover review on a two-furnace, six-line plant running a Heye International IS machine on mid-2000s servo timing controls. Preheat oven discipline was inconsistent across shifts. The morning shift soaked to spec. The afternoon shift pulled moulds at T+2h20min because the job change was running behind. First-ware reject rates between those two shifts on the same SKU were sitting at a 34% gap in a rolling four-week comparison. Same machine. Same glass. Same mould set. Different soak time.

Not a forming problem. A preheat discipline problem.

Why most plants keep under-preheating

Look, the data says one thing and the floor says another. Operators know from experience that moulds feel hot enough at around 350-380°C. The palm test stops being useful above about 250°C, but the instinct persists because nobody has ever quantified what a 60°C shortfall in preheat surface temperature costs the plant in first-ware rejects and time-to-stable-pack. Plants that don't measure preheat surface temperature can't make that case to a shift team under pressure.

Schedule compression is the other driver. A job change running 30 minutes behind will almost always sacrifice soak time, because soak is the only step in the changeover sequence that feels passive. It's just waiting. So it's the first thing compressed when the line manager is on the phone asking when the new SKU is running.

And this is where cross-shift variance compounds the problem. If your afternoon superintendent holds the full soak and your night superintendent pulls moulds early because his line was 20 minutes late coming down, you'll never get a clean comparison in your first-ware data. The variable isn't visible because nobody's measuring preheat surface temperature at the point the moulds leave the oven. They're measuring how long the oven was on. That is not the same number.

You cannot manage a preheat strategy you haven't measured. Oven-on time is not mould surface temperature. One tells you how long the oven ran. The other tells you whether your glass is safe.

Making preheat discipline hold when the schedule is against you

The fix is straightforward to describe. Every mould set needs a preheat record card with four entries:

• Target surface temperature (480°C ±10°C, checked by contact thermocouple at the blank mould body, not the base plate)
• Ramp profile: phase one at 80-100°C/h to 420°C hold, phase two at 40°C/h to final target
• Soak start time and confirmed soak end time
• Sign-off name and timestamp

The hot-end superintendent owns that sign-off. Not the operator, not the mould shop, not the shift engineer. If the line is behind schedule, that sign-off is the control gate. The schedule adjusts. The spec does not.

That sounds obvious. In practice it isn't. I've been in container glass plants across the GCC running mixed OEM fleets on multiple furnaces where nobody owned preheat sign-off formally. The mould shop thought it was the operator's job. The operator assumed the superintendent had already checked. That unowned handover between mould shop and hot end is endemic across the industry, and it's fixable with one named step and one signature.

If you're rebuilding your job change process from the ground up, a vendor-neutral forming audit at Lean Glass starts exactly here: establishing ownership of each step in the changeover sequence before touching any equipment settings. That's what a container glass consultant finds in the first hour of a forming engagement. If cross-shift variance is your visible symptom, the Job Change Tool is where you lock preheat specs alongside every forming parameter so they stop living in a shift fitter's notebook. And if you suspect the preheat oven itself is the problem, a hot end audit will tell you whether it's delivering consistent temperature across the mould set or running a hot-zone bias that no soak time will fix.

Preheat is a process. It has a target, a ramp, a soak, a measurement point, and an owner. Treat it as anything less and you're converting thermal discipline into first-ware rejects on every single job change.

So what does your preheat card look like right now?