Why moulding rejections keep returning: what job cards miss

One card reads 340 rejected of 5,000. The supervisor blames a bad shift. The manager, reading it a week later, cannot tell whether the mould, the press, the dryer or the resin lot was at fault.

Next week the same housing returns on another press, with the same short shots at the same far corner of the part, away from the gate. Nothing was written down that could separate one cause from another, so the pattern repeats on every press and every shift for months. Scrap is the symptom.

The log must carry enough fields that a single query points to the mould, the resin lot, the dryer or the setting that changed.

Start with where the part was made, namely the mould number, the cavity number, the machine number and the shift that was running at the time. Add what went into it: resin grade, lot number, regrind ratio and, for nylon or PET, the drying time and dryer dew point. Then record what the process did, namely cycle time against the standard, melt and mould temperature, hold pressure and shots since the last mould maintenance.

Each rejected piece needs a defect code. Short shot, flash, sink mark, burn mark, warpage and splay each point to a different cause, from a worn vent to wet resin to low clamp force. One bucket hides all of them.

Use a tablet at the press. Let the operator pick the mould, cavity and defect from a short list, because free text becomes noise within a week and nobody can group it.

Group rejections by cavity and one cavity usually stands out, because a worn vent or a damaged gate in cavity 6 produces short shots there and nowhere else. Without that field, the toolroom polishes the whole mould, and cavities that were running well lose their sharp edges for nothing at all.

Group by resin lot and flash may cluster after a lot change, because the melt flow index of the new lot differs from the old one. Group by regrind ratio and splay may rise when the dryer is not holding moisture down, which points at the dryer rather than the operator.

Group by shots since maintenance and flash climbs as parting lines wear, which gives you a curve that shows when the mould should come off the press. Pull it before the customer complains.

Operators stretch the cycle to reduce rejections and then forget to shorten it, so the job looks fine on quality while it quietly loses output every hour. Log actual cycle time against standard.

A setter who raises barrel temperature or hold pressure to fill a worn mould trades short shots for flash or burn marks, and the card never says so. Log each change with reason and name.

Same setting adjusted on every run? When the log shows that, the mould needs repair, not a better operator or another training session for the night shift on that press.

Link the rejection log to the mould history, so every repair, polish and vent cleaning sits on the mould’s card with its date and shot count. When flash returns soon after a polish, you can see that the repair held for only a few thousand shots and decide whether the steel needs rework.

Trigger maintenance by shot count, not by calendar, because a mould that ran a million shots needs attention sooner than one that sat idle on the rack.

Start small. Pick your three worst moulds, capture the fields above for a month and draw one Pareto chart of rejections by cavity, which usually pays for the effort.