Injection Molding Cpk Requirements

Why does a multi-cavity tool almost always misrepresent capability when cavities are pooled, and how should the study actually be structured?

Use the customer requirement, drawing, validation protocol, and cavity-specific Control Plan. Common program examples—not universal requirements—may include:

• Cpk ≥ 1.67 for some launch or approval studies per cavity when specified.
• Cpk ≥ 1.33 for some ongoing studies per cavity when specified.
• Cavity-balance limits must come from the validated process and part requirements; do not assume one universal percentage band.

Injection-molding capability is a per-cavity property. A multi-cavity or family tool is not one process — it is N processes sharing a machine and a melt. Pooling cavities inflates spread and hides the offending cavity; splitting cavities exposes the real state. Customer requirements (automotive, medical, packaging) drive the target; injection molding itself has no universal capability standard.

1. Design the study around cavity ID as a required tag — no untagged part enters the dataset.
2. Verify mold and melt temperature stability before sampling.
3. Verify material lot, moisture and regrind ratio; record on the study.
4. Sample after the machine and tool are in steady state, not immediately after start-up.
5. Compute per-cavity Ppk for PPAP; report the worst cavity, not the average.
6. Run a cavity-balance check (short shot or weight study) at least once per PPAP.
7. Update the Control Plan with per-cavity reaction rules, not tool-level rules.

• Cavities pooled into one Ppk; failing cavity hidden.
• Regrind ratio changed between the PPAP run and serial production.
• Dryer dew-point not recorded; hygroscopic resin variation blamed on machine.
• Cavity balance drifted after hot-runner PM; not re-verified.
• Cooling circuit swapped for maintenance; shrinkage shifted; capability unchanged in report.
• Family mold PPAP submitted with one part number's data extrapolated to others.

• Pooled multi-cavity Ppk is the single most common way capability studies pass on paper and fail on shipment.
• Material lot changes cause more Cpk drops than any tool problem; require lot traceability on every study.
• Family molds should almost never share one capability report — each part number is its own study.
• A cavity that "was fine last month" is usually a gate-wear or hot-runner drift, not a random excursion.

• Do not pool cavities to reach a target Ppk — it is a misrepresentation of the process.
• Do not compute capability across a material-lot change without segmenting.
• Do not use capability from a technician-optimized set-up window as evidence of production capability.

• Per-cavity Cpk vs pooled Cpk: The gap between them quantifies cavity imbalance.
• Cpk vs shrinkage variation: Long-term Cpk decay is usually a shrinkage / material-lot signal.
• Capability vs process window: Cpk is only valid inside the validated melt / mold / cooling window.

Use the free Process Capability Calculator to compute Cp, Cpk, Pp, Ppk, sigma level, and predicted PPM from your measurement data — no sign-up required.

• Never accept a molding capability study without cavity tags on every measurement.
• Never claim capability outside the validated process window — window first, capability second.
• Never treat regrind ratio as a free variable — lock it into the Control Plan.

• ✓Every measurement tagged with cavity ID
• ✓Per-cavity Ppk reported; worst cavity is the acceptance value
• ✓Cavity balance verified (short shot or weight)
• ✓Material lot, moisture and regrind ratio recorded
• ✓Mold, melt and cooling within validated window
• ✓Family-mold parts studied individually, not extrapolated
• ✓Control Plan reaction rules defined per cavity

• Process condition: statistical stability is required.
• Distribution assumption: use a distribution model justified for the data.
• Confirm process stability and measurement-system adequacy before interpreting a capability index.
• Use a justified distribution model or non-normal method when the normal model is unsuitable.
• Numerical targets shown on industry pages are common examples, not universal requirements. The contract, drawing, customer-specific requirement, Control Plan, and validation protocol take precedence.