Statistical Process Control Application in Multi Cavity Injection Tooling Qualification Protocols

Core

First shots from a thirty-two cavity injection mould reflect the physical specifics of each cavity location. Dumping parts from every cavity into a single tote to calculate overall capability distorts the statistics, producing an unnaturally wide bell curve and depressed capability indices that tell a toolmaker virtually nothing. Qualifying a multi-cavity tool requires separating spatial variation across the mould layout from temporal variation over press operating hours.

Each cavity acts as an independent production unit linked only by the runner system and clamp pressure. Dimensional differences between impressions stem from CNC toolpath offsets, wire Electrical Discharge Machining drift, uneven heat-treatment hardening, or local polishing variations on core inserts. Even when tools are built to nominal CAD models, steel dimensions across cavities still vary within a five to ten micrometer band.

If individual cavities produce tight distributions around slightly different means, pooling the parts creates a synthetic normal distribution with an inflated standard deviation.

A multi-cavity tool operates as multiple distinct production units sharing a single hydraulic cycle.

Adjusting press settings based on pooled data inevitably runs into dead ends. Increasing pack pressure to bring a tight cavity into spec forces an adjacent cavity to flash, while trimming two seconds of cooling time warps parts in the block’s hottest quadrant. Qualification has to isolate steel geometry before machine stability metrics are brought into the analysis.

Inspecting the mould steel directly confirms alignment before resin ever hits the barrel. Toolmakers log coordinate measuring machine measurements of critical core pins, slide travel limits, and insert pockets on the inspection ledger. Initial dry-cycling verifies mechanical movement, platen parallelism at fifty percent clamp tonnage, and ejector alignment.

Polymer injection starts only after static steel metrology confirms cavity dimensions fall within thirty percent of the total drawing tolerance band.

Mould cavities exhibit characteristic mechanical deviations during actual injection cycles:

• Core deflection under injection pressure pushes internal diameters toward the parting line when asymmetric gating creates uneven side-loading during fill.
• Thermal expansion differential across plates shifts outer cavities outward relative to the central sprue, altering part pitch across long cavity rows.
• Gate orifice variation causes uneven filling rates when wire EDM wear or manual deburring leaves entrance areas differing by over five percent across impressions.
• Venting clearance collapse traps gas in outer cavity corners, preventing full packing and driving localized shrinkage spikes.

Signing off on multi-cavity tooling without tracking cavity identity hides localized dimensional defects inside lot averages, driving up scrap once automated assembly lines jam in production.

Subgroup

How data is collected determines whether control charts signal real press instability or simply map cavity layout geography. Rational subgrouping relies on maintaining explicit cavity identity through measurement, logging, and analysis. Measuring five parts pulled at random from an eight-cavity tool fails to capture true within-subgroup variation over time; it merely reflects the random selection of cavities, skewing X-bar and R limits.

Structuring rational subgroups for multi-cavity tools typically follows two paths. One option tracks every cavity as its own control stream, generating thirty-two separate X-bar and S charts for a thirty-two cavity mould. While metrology-intensive, this pinpointed tracking shows immediately which insert is wearing or drifting.

The alternative uses nested analysis of variance within each shot, treating cycle time as the main subgroup and individual cavities as nested factors within that cycle.

Four consecutive shots measured across every impression establish the baseline variance structure of a new mould.

During operational qualification, as tracking cushion drop reveals potential check ring leakage, technicians collect full shots across twenty consecutive cycles under steady-state conditions. Measuring critical dimensions across every cavity in those twenty cycles yields four hundred data points on a twenty-cavity tool, providing the structured dataset needed to partition within-cavity, cavity-to-cavity, and shot-to-shot variance through nested analysis.

Where variance sits determines how to fix the problem. If cavity-to-cavity variance accounts for over sixty percent of total variation, no amount of press tuning will balance the parts; the toolmaker must adjust the steel. When shot-to-shot variance dominates, the press itself lacks pressure stability, melt homogeneity, or screw recovery consistency.

Qualification protocols set clear thresholds to distinguish machine instability from tool defects.

1. Execute consecutive shot collection by running validated cycle parameters for sixty minutes, purging thirty initial cycles, and saving five consecutive complete shots in indexed trays by cavity location.
2. Conduct dimensional measurement serialization using CMM routines or optical profile scanners, logging three replicate measurements per critical feature to isolate measurement error before evaluating parts.
3. Compute variance partitioning coefficients using nested linear models to compare spatial tooling variance against temporal process variance, checking steel consistency across all impressions.
4. Establish cavity control limits by setting individual three-sigma control bounds for each cavity location rather than applying pooled control bands across distinct impressions.

Contracts specifying ISO 22514-7 compliance require process qualification dossiers to include variance partitioning before granting final production part approval.

Balance

Molten polymer behaves as a non-Newtonian fluid whose apparent viscosity drops under high shear rates. Even in naturally balanced runner layouts where flow paths look identical on CAD drawings, shear heating along channel walls creates thermal differences between inner and outer cavities. Melt running through the center of the channel stays cooler than polymer sheared along the walls, feeding streams of different temperatures into individual impressions.

Because viscosity varies across resin lots, an eight degree Celsius temperature rise along a runner branch reduces flow resistance enough to make the warmer cavity fill faster and pack tighter than adjacent ones. To quantify fill balance, technicians turn off hold pressure, trim injection stroke to fill parts to eighty percent volume, and collect short-shot samples. Weighing these incomplete parts exposes the true hydrodynamic balance of the runner manifold.

Dynamic filling imbalance exceeding five percent across cavity weights prevents simultaneous gate freeze across the tool.

Determining filling balance percentage is straightforward: divide the weight of the lightest short shot by that of the heaviest short shot from the same cycle and multiply by one hundred. Medical and precision technical mouldings generally require a balance above ninety-five percent, while high-cavitation thin-wall tooling often targets ninety-eight percent. Below ninety percent, no amount of barrel pressure adjustment will yield uniform part dimensions.

Temperature variations across the platen add another layer of complexity. Large tool blocks develop thermal gradients between hot central manifold areas and cooler outer edges. Cooling circuits wired in series pick up heat as water flows from cavity one through cavity sixteen; water entering at eighteen degrees Celsius can reach twenty-six degrees Celsius by the outlet on long serpentine runs.

This temperature rise skews cooling rates and shrinkage in downstream cavities, driving systematic dimensional drift across the mold face.

Adjusting barrel heat and packing pressure during production setup does not resolve underlying cavity-to-cavity weight imbalances.

Index

Evaluating multi-cavity tooling relies on capability metrics defined in ISO 21747 and DIN 16742. Pooling data to calculate potential capability (Cp) or performance capability (Cpk) causes misleading results, flagging capable tools as failures or concealing out-of-spec cavities inside acceptable averages. For instance, if thirty-one cavities achieve a Cpk of 1.67 while cavity seven drops to 0.72 because of a misaligned ejector sleeve, the pooled data can still show a passing Cpk of 1.35, obscuring twenty thousand non-conforming parts per million from that single cavity.

Qualification protocols require evaluating capability indices for each cavity individually, generating a Cpk matrix across the entire mould layout. The qualifying metric for the tool is the lowest individual cavity Cpk rather than the average. If any single impression falls below the contractual limit ~ typically 1.33 for standard parts or 1.67 for safety-critical dimensions ~ the tool fails qualification.

Contractual qualification clauses under DIN 16742 require every individual cavity to achieve the target Cpk value independently.

Long-term performance indices (Pp and Ppk) capture process variation across extended runs that encounter material lot shifts, regrind additions, ambient temperature changes, and crew handovers. Short-term machine capability (Cmk) is measured during tool trials using fifty consecutive shots per cavity with fixed press settings, isolating mechanical repeatability from environmental noise.

Qualifying high-cavitation tools with ninety-six or one hundred twenty-eight cavities creates major metrology bottlenecks. Measuring every cavity across thirty production shifts produces nearly two hundred thousand data points for just one critical feature. To manage inspection costs, engineers often use extreme-cavity sampling: all cavities are measured in five initial shots to identify the statistical extremes (the upper, lower, and highest-variance positions).

Production monitoring then tracks these sentinel cavities, working on the premise that as long as the extremes remain in spec, interior cavities are compliant.

Whether extreme-cavity sampling remains statistically reliable over long production runs with abrasive glass-filled resins, which erode gates unevenly, remains an open question.

Release

Steel adjustments are the final physical step in tool qualification. When nested SPC shows an individual cavity is capable but off-target, toolmakers perform controlled, steel-safe modifications. Designers intentionally leave excess stock ~ making internal cores oversized and external features undersized ~ so metal can be removed during tuning.

If a cavity consistently molds external bosses two hundredths of a millimeter undersize, the cavity pocket is deepened using precision sinker EDM.

Modifying steel before establishing a stable process creates major risks. If metal is cut to center a dimension while the press runs with an unstable cushion or incomplete packing, the dimension will drift again once operating parameters are corrected. Steel modifications should occur only after operational qualification is confirmed by cavity or runner pressure transducers showing consistent pressure transfer during hold.

Final capital release depends on signed qualification dossiers containing full statistical proof. Buyers typically withhold thirty to fifty percent of milestone payments until the mould meets both capability targets and cycle time requirements. Qualifying a sixteen-cavity connector tool at an eighteen-second cycle to avoid flash when the contract specifies twelve seconds constitutes a breach, as cycle time dictates unit economics over the life of the tool.

Once statistical evidence confirms all cavities meet capability targets at the quoted cycle time, the team runs formal factory acceptance. This test involves four to eight hours of continuous, unassisted press operation with zero critical defects and stable control charts across every cavity stream. Signing the acceptance document transfers ownership to the plant and commits the supplier to maintaining these capability levels in serial production.

Tool steel should never be cut until three consecutive production runs confirm consistent cavity mean offsets.