Pre Production Mold Pilot Run Sampling

• 1
  What pre production mold pilot run sampling actually measures
• 2
  Which process windows the run has to hold
• 3
  When a pilot run cannot tell you the truth
• 4
  What the sampling report has to contain
• 5
  Why tooling and machining should sit in one process
• 6
  Pilot run readiness: when to sample and when to wait

The tool is finished, the steel is cut, and the water lines are drilled. Pre production mold pilot run sampling is the point where the tool stops being a promise and starts producing data. The run uses the final cavity and core, the intended process parameters, and the material the customer will buy in volume. Output is usually dozens to a few hundred parts, not a full production order.

A pilot run is not a dimensional check with a nicer name. It tests the whole chain: fill balance across cavities, cooling efficiency, ejection dynamics, gate freeze-off, and how the part behaves after the mold opens. If the tool has four cavities and one fills 40 ms late, you will see it here, not in a first-article report on a single cavity.

The sample set is what makes the data usable. Ten parts prove a tool can make a good part once. A statistically meaningful batch proves the process can repeat. That is the difference between a toolroom demo and a production process you can hand to a quality system.

Downstream operations belong in the same run. If the part will be CNC machined on a fixture after molding or casting, hold the fixture and machine a few pilot parts. The molded feature tolerance and the machined feature tolerance stack, and that stack is where most late surprises live.

• 1
  Final tool onlyOff-tool or soft-tool samples do not count.
• 2
  Production settingsMold temperature, hold pressure, cooling time as intended.
• 3
  Full metrologyNot a spot check on two or three dimensions.
• 4
  Downstream includedMachine and finish the pilot parts the same way as production.

A pilot run that sits at the center of the window proves the tool works at one point. That is not enough. Run the tool at the low and high edges of the intended window too: minimum and maximum mold temperature, short and long hold, low and high injection speed. If the part stays in tolerance across the window, the process has room to drift in production.

For machined features, the equivalent is cutting parameters. On aluminium 6061-T6 and 7075, finishing passes at Ra 0.8–1.6 μm hold well if the tool path and coolant are right. Push the feed and the finish moves to Ra 1.6–3.2 μm. The pilot run is where you find out which side of that line your process sits on.

Tolerance is a system property, not a tool property. A tool cut to ±0.005 mm can still produce parts that drift when the coolant loop is undersized or the press runs hot. Pilot sampling with a thermal log catches that before a production order is committed.

Write the window down. The pilot run report should state the parameter range that produced conforming parts, not just the single set point used on the day. That range is what operators use six months later when a new lot of resin arrives.

• 1
  Edge runsLow and high limits, not just the nominal set point.
• 2
  Thermal loggingRecord mold and melt temperature through the run.
• 3
  Stack-up checkMeasure molded and machined features together.
• 4
  Written windowParameter range goes into the report, not just a set point.

A pilot run on a tool that is not final is waste. If the cavity is still going to change, or the gate location is under review, the sampling data describes a tool that will not exist. Fix the design first. Sampling is a verification step, not a design tool.

Single-cavity trials on a multi-cavity tool also mislead. One cavity filling well says nothing about the balance across eight. If the runner is not balanced, the pilot run will show it, but only if all cavities are measured, not just the one that looks good.

Material substitution breaks the result. A pilot run in a surrogate resin does not transfer to the production grade. Shear sensitivity, shrinkage, and cooling rate all shift. If the production material is not available, say so and label the data as indicative, not as process validation.

Small sample sizes hide drift. Twenty parts can look fine while the process is walking out of tolerance. If the feature is critical, run enough parts to see the trend, or run the tool twice on separate days and compare.

• 1
  Not final designSampling a tool that will change proves nothing.
• 2
  Single cavity onlyMulti-cavity balance stays unknown.
• 3
  Surrogate materialData does not transfer to the production grade.
• 4
  Too few partsDrift stays invisible; run across days.

A pilot run without documentation is a story, not evidence. The report needs the parameter window, the measurement plan, the raw data, and the pass or fail call against the print. It also needs the material lot, the machine, and the date. Six months later, that is the only way to know whether a shift in production is the tool or the material.

Inspection coverage matters as much as the numbers. Spot checks on two dimensions can miss a warped part. A full layout on a few parts plus in-process monitoring on the rest is usually the practical balance. State which dimensions get full layout and which get monitored, and why.

For regulated work, the report feeds the quality system. IATF 16949 and ISO 13485 audits look for evidence that the process was validated, not just that parts passed. A pilot run report with raw data and a signed acceptance is that evidence.

Keep the rejected parts too. The parts that failed at the edge of the window tell you more about the process than the ones that passed. Store them with the report so the next engineer can see the failure mode.

• 1
  Parameter windowRange that produced conforming parts.
• 2
  Raw dataMeasurements, not just pass or fail.
• 3
  Acceptance callSigned against the print and spec.
• 4
  Keep failuresEdge-of-window rejects document the boundary.

Most pilot runs fail at the handoff. The tool shop samples the molded part, the machine shop gets a different drawing revision, and the two sets of measurements never meet. When tooling and CNC machining sit in one process, the stack-up is measured once, on the same parts, against the same revision.

At GreatLight we run 127 high-precision CNC machines, including 16 simultaneous 5-axis machining centers, across three wholly-owned plants. When a pilot run needs post-machining, the fixture and the cutting parameters are set during the same run, not after the tool is shipped.

Metrology closes the loop. A full layout on the pilot batch, plus in-process monitoring, shows whether the tool or the machining is moving. Tolerance is ±0.005 mm and finish down to Ra 0.2–0.8 μm when the drawing calls for it, and the pilot run is where those numbers get proven on real parts.

The pilot run also feeds DFM. If a feature is hard to mold and hard to machine, that shows up in the samples. Changing the feature at pilot stage costs a drawing revision. Changing it after production starts costs a tool.

• 1
  One revisionTool shop and machine shop work from the same drawing.
• 2
  Fixture at pilotMachining set up during the run, not after.
• 3
  Metrology loopFull layout plus in-process monitoring.
• 4
  DFM feedbackHard features surface while changes are cheap.