A technical moulder’s nylon gear quality depended on which shift was running. Scientific moulding studies and one locked process sheet per tool cut weight variation by 79%, dimensional rejects from 2.8% to 0.4% and setter adjustments from 11 to 2 per shift.
Key numbers
At a technical moulder producing nylon gears, the quality manager could tell which shift had made a batch without looking at the label. Day shift parts were usually fine. Afternoon parts were more variable. Night shift parts were the worst. Each setter had a preferred way of running the process, and each was confident theirs was correct.
The situation
Gears are unforgiving parts: small dimensional changes affect mesh, noise and wear. Dimensional rejects averaged 2.8%, and setters were making around 11 adjustments per shift trying to keep parts in tolerance, each one changing the process a little further from where it started.
What the data showed
Part weight is a sensitive indicator of process consistency, so the team weighed samples from every shift over several weeks.
The standard deviation of part weight was 0.31 g on day shift, 0.44 g on afternoons and 0.52 g on nights. Process records showed why: each setter used different injection speeds, switchover points and hold settings, and adjusted them frequently. The process had never been defined by data, so it was defined by people, exactly the situation described in scientific moulding: replacing guesswork with data.
What changed
- Viscosity studies on each tool to choose injection speeds where normal material variation had little effect on fill.
- Cavity balance studies to confirm every cavity filled consistently.
- Gate freeze studies to set hold time just beyond the point where the gate froze.
- A decoupled process: fill to around 98% under velocity control, switch over by screw position, then pack and hold under pressure control.
- A single documented process sheet per tool, locked on the machine controller, with the reasons for each setting recorded.
- Setters trained in scientific moulding principles, so they understood the process window and why it worked.
Results
Weight variation fell by 79% and became consistent across shifts, at 0.08–0.09 g on every shift. Dimensional rejects dropped from 2.8% to 0.4%. Setter adjustments fell from about 11 per shift to two, because the process no longer needed constant correction.
Why it lasted
The process sheet explained why each setting was chosen, so setters could see that changing it would make things worse, not better. When a genuine problem occurred, they had a known good process to compare against, which made troubleshooting faster.
Lessons for your operation
- Weigh parts by shift. It reveals process inconsistency quickly and cheaply.
- Define the process with data, not preference: viscosity, balance and gate freeze studies.
- Use machine-independent settings and document the reasons behind them.
- Train setters in the principles, not just the settings. Understanding is what stops drift.
