Dimensional drift in polyethylene articles between a trial order and full production is not explained by a single data-sheet property. The problem arises from the interaction among semicrystalline morphology, feedstock heterogeneity, plasticating conditions, and cavity-pressure history. Polyethylene grades span a density range of 0.910 g/cm³ to 0.970 g/cm³ under ASTM D792-20, with low-density, linear low-density, high-density, and ultra-high-molecular-weight types exhibiting different crystallization onset temperatures and volumetric shrinkage. A trial lot is commonly drawn from a single homogenized batch and processed on one machine with a narrow shot-to-shot weight window; a full production campaign combines multiple lots, often blends regrind, and subjects the melt to a larger plasticating unit with longer residence time. These shifts do not require an off-specification resin to produce measurable dimensional change.

The relevant measurement chain begins with ISO 1133-1:2022 for melt mass-flow rate, ASTM D1238-20 for melt index, ISO 1183-1:2019 for density, and ISO 294-4:2018 for shrinkage of moulded test specimens. A trial-order verification that records only tensile yield and density will not capture the melt-flow drift that changes packing transmission. Production-scale dimensional control requires cavity-pressure measurement at the gate and post-gate locations because polyethylene continues to crystallize and contract after demolding; residual shrinkage can continue for 24 h to 48 h depending on wall thickness and cooling rate. Published data for this specific configuration is limited, but the mechanisms are measurable with standard rheological and thermal instrumentation.

Why Polyethylene Parts Drift When a Trial Order Goes to Full Production

The drift is not a single phenomenon. In high-density polyethylene injection moulding, it is the accumulation of several independent variables: low glass-transition temperature, high crystalline fraction, and slow post-mold crystallization. The crystalline fraction of HDPE can range from 60% to 80%. As the part cools from a melt temperature of 190 °C to 220 °C to a mould temperature of 20 °C to 40 °C, the crystallization exotherm releases enthalpy and reduces specific volume. Any change in cooling rate, packing pressure, or nucleating agent concentration shifts the lamellar thickness distribution. A production tool may have a different surface temperature map than the trial tool because of higher coolant loads, scale-up of water circuits, or cycle-time pressure. These thermal differences change the solidification front velocity and therefore the anisotropy of shrinkage.

Representative variables that differ between trial and full production are summarized below; these are industrial operating ranges against which dimensional drift is commonly investigated.

Melt Flow Rate, Molecular Weight Distribution, and Packing Pressure Decay

Melt mass-flow rate is measured at 190 °C under 2.16 kg; the test gives a single viscosity point and cannot resolve the high-molecular-weight tail. In HDPE blow-moulding and injection grades, gel-permeation chromatography typically reports a polydispersity index between 4 and 12 depending on catalyst system. A full production lot blended from two reactors or a different catalyst donor may carry the same nominal MFR but a different polydispersity index. This shifts the pressure gradient along the flow path. The pressure at the gate is dissipated through the sprue, runner, and cavity; low-molecular-weight chains lower viscosity at high shear and allow a deeper penetration of packing pressure before gate freeze. If gate freeze time changes from 8 s to 6 s, the effective holding time is shortened even when the machine setting is unchanged. The resulting mass difference may be mistaken for a mould leak or screw wear when it is actually a resin architecture effect.

The MFR difference of 1.2 g/10 min may remain within the supplier’s release tolerance, particularly for broad-specification high-density grades. However, MFR is a low-shear rheological indicator; it does not capture the high-shear viscosity relevant to injection filling. A broader molecular-weight distribution can improve shear thinning while altering die swell and melt elasticity. On a full production line, this change is expressed as a different switchover point from velocity control to pressure control. If the switchover position is fixed by screw displacement, a lower-viscosity lot transfers more material before gate freeze, producing a heavier and less-shrunken part; a higher-viscosity lot underfills the same displacement and increases sink. The cavity-pressure transducer therefore becomes the only reliable process signal for holding-pressure adjustment. Injection moulders routinely observe part mass shifts of 0.2% to 1.0% with co-approved MFR variations.

A second production-scale source is screw and barrel geometry. Trial moulding may be performed on a 30 mm diameter screw with a 25:1 L/D ratio and a shot weight below 30% of barrel capacity. Full production may transfer to a 65 mm screw with a 22:1 L/D ratio and a larger shot capacity. The two machines do not apply the same shear history even when the barrel set-points are identical. In a larger screw, pellet melting occurs over a longer path but the surface renewal rate is lower, and the melt may spend 3 min to 8 min in the barrel. HDPE begins to oxidize above 240 °C when oxygen is present, producing a crosslinked gel fraction and a measurable shift in die swell. The effect on part dimensions is indirect but real: gel particles interfere with the packing flow and produce local sink or warpage. The solution is to map barrel temperatures with a melt temperature probe rather than trusting set-point values.

Tool Temperature Differentials Reshape Semicrystalline Morphology

Mould temperature is the strongest process variable controlling polyethylene shrinkage after molecular architecture. In a semicrystalline polymer, shrinkage is a combination of thermal contraction and crystallization. Rapid cooling produces a thinner crystalline lamellar structure and lower initial degree of crystallinity; slower cooling permits more extensive lamellar growth and higher volumetric shrinkage. If the trial tool maintained a cavity surface temperature of 30 °C and the production tool operates with a coolant inlet temperature of 15 °C to 18 °C, the cooling rate increases and the part may shrink less in the flow direction but more in the transverse direction because of frozen-in orientation. Temperature differences between fixed and moving halves of more than 5 °C are sufficient to produce warpage in a flat HDPE cover. Full production also increases the heat load on the tool; water-line scaling or insufficient flow can raise local cavity temperatures by 8 °C to 12 °C, changing the solidification front.

The standardized shrinkage measurement in ISO 294-4:2018 uses a 60 mm × 60 mm × 2 mm plaque or specified specimen depending on mould design. Shrinkage is measured at 24 h after demolding and expressed as a percentage difference from the cavity dimension. For HDPE, parallel-flow shrinkage is typically 1.5% to 3.0%, and perpendicular-flow shrinkage can be 0.5% to 1.5%, but these values depend on packing pressure and cooling rate. A production lot that shows the same 24 h shrinkage may still exhibit later dimensional shift between 48 h and 168 h because secondary crystallization and physical aging proceed at different rates for different cooling histories. Post-mold annealing at 80 °C for 1 h is sometimes used to stabilize dimensions, but it increases shrinkage by advancing crystallization.

After 15 wt% Regrind, the Melt-Index Shift Becomes a Dimensional Variable

Trial orders usually consume virgin pellets; full production almost always includes regrind because sprues, runners, and rejected parts are recycled. The regrind fraction is not inert. Each heat history imposes chain scission or branching, and the particle size of ground feedstock changes melting behaviour. A regrind fraction of 15 wt% can shift the melt mass-flow rate by 0.5 g/10 min to 2 g/10 min, depending on the number of heat cycles and the antioxidant package. The dimensional consequence is not uniform: higher melt flow improves penetration into thin walls but reduces melt-cushion consistency. The screw also compacts flake regrind differently from cylindrical pellets, altering feed-batch density and melt pressure. If the molder holds first-stage injection pressure constant, the switchover cushion becomes variable and the holding phase starts from a different compressibility state.

Oxidative stability is relevant: HDPE resin stabilized with a blend of hindered phenolic antioxidant and phosphite carries an oxidation induction time under ASTM D3895 or ISO 11357-6. Multiple extrusion cycles consume the antioxidant package. When the oxidation induction time falls below 20 min at 200 °C, the regrind may begin to generate gel particles that disturb packing. The dimensional stability protocol is therefore required to limit regrind to 15 wt% unless the tolerance allows a wider process window, and to verify the regrind MFR separately rather than averaging it with virgin pellets.

Masterbatch addition can also shift dimensions between trial and production. A calcium carbonate masterbatch at 20 wt% loading increases density and reduces linear shrinkage while increasing part weight. If the trial was run with natural resin and production introduces a colour or slip masterbatch, the solidification rate changes because the filler acts as a nucleating agent and modifies thermal conductivity. Talc-filled PE can reduce shrinkage anisotropy relative to unfilled PE, but the melt viscosity rises and the required packing pressure grows. The molder must re-establish the holding-pressure window rather than copying the trial parameters.

The production acceptance protocol requires closure of the measurement loop. A cavity-pressure trace is recorded at the gate and at the end-of-fill position for every lot; part mass is weighed on a balance with 0.01 g resolution; dimensional measurements are referenced to the mould-specific shrinkage map rather than to the trial part drawing alone. When the relationship between MFR, gate freeze time, cavity pressure, and post-mold shrinkage is recorded continuously, a lot-to-lot MFR change of 0.8 g/10 min is identified as a deterministic change in packing and crystallization rather than random quality drift.