Measuring PBT Shrinkage Properly: Flow vs Cross-Flow and Cycle-to-Cycle Spread starts with the recognition that a single shrinkage value on a polybutylene terephthalate datasheet is not sufficient for mould validation or process-capability assessment. Polybutylene terephthalate is a semicrystalline polyester that exhibits anisotropic moulding shrinkage: flow-direction contraction is routinely lower than cross-flow contraction in the same specimen. The offset is not a measurement anomaly but is produced by molecular orientation, fibre orientation in reinforced grades, crystallinity gradients through the wall thickness, and in-plane constraint imposed by the gate and cavity walls. A proper measurement therefore records paired flow and cross-flow values from the same plaque, the cavity pressure history, the elapsed time after demoulding, the conditioning atmosphere, and a dispersion estimate across at least 10 consecutive cycles.

The term flow direction in shrinkage specifications denotes the axis from the gate along the advancing melt front; cross-flow denotes the orthogonal in-plane axis. The distinction is operational, not merely geometric. Cavity pressure at the gate and cavity pressure at the last point of fill differ, and this gradient controls local packing. A single PBT plaque can therefore show different shrinkage behaviour if only one axis is sampled. That is why the data package is not complete until the flow axis, cross-flow axis, and the gate position are documented together.

What Introduces the Flow-to-Cross-Flow Offset in PBT Shrinkage Measurements?

During filling and packing, PBT macromolecules and, in filled compounds, glass fibres orient along the melt-flow streamlines. Flow-direction orientation reduces linear contraction because oriented chains and fibres act as high-modulus constraints along the orientation axis. Perpendicular to that axis, relaxation of the matrix is less constrained, and cross-flow contraction therefore remains larger. On an edge-gated 60 mm × 60 mm × 2 mm plaque, unfilled PBT typically produces flow-direction shrinkage of 1.4 % to 2.0 % and cross-flow shrinkage of 1.7 % to 2.5 %. A 30 % glass-fibre reinforced PBT compound often reduces flow-direction shrinkage to 0.3 % to 0.6 % while cross-flow shrinkage remains at 0.7 % to 1.1 %. The anisotropy ratio is therefore grade-dependent, and using only the flow-direction value can understate the cross-flow dimension change that a mould must accommodate.

The measured anisotropy is influenced by cavity thickness, gate geometry, and mould temperature. When mould surface temperature is increased from 60 °C to 90 °C, the cross-flow-to-flow ratio often narrows because of delayed vitrification and more uniform crystallisation. Under ISO 294-4 and ASTM D955, shrinkage is calculated from the cavity dimension and the conditioned moulding dimension as S = (L_c − L_m) / L_c × 100 %; the same calculation must be applied separately to flow and cross-flow axes, with the flow axis identified from the gate position and melt-flow front geometry. If the plaque is centre-gated and radial, reporting a single circumferential value without angular documentation obscures the directional dependence.

At 30 % glass-fibre loading, the cross-flow:flow ratio can reach 1.5–2.5; at 50 % glass-fibre loading the ratio may be higher, but absolute cross-flow shrinkage can be lower because the occupied volume fraction of non-shrinking fibre limits all dimensional contraction. The same logic applies to mineral-filled PBT grades, although platy fillers may reduce cross-flow shrinkage more effectively than spherical fillers. Published data for high-filler systems remains grade-dependent and should be derived from the actual mixed compound rather than extrapolated from resin-only data.

Cycle-to-cycle spread in PBT is controlled less by the resin than by holding-pressure repeatability, check-ring seating, screw recovery consistency, and cavity-temperature drift. A production moulding trial that records only the arithmetic mean of 5 shots cannot detect a systematic shift of 0.05 % caused by check-ring leakage. The cycle-to-cycle dispersion should be reported as a ±3s interval or a coefficient of variation because shrinkage data on PBT often contain a skewed tail toward lower flow shrinkage when the holding phase is shortened by premature gate freeze-off. Paired flow and cross-flow measurements are required; a defect that changes cross-flow contraction while leaving flow contraction apparently stable can pass undetected when only the flow axis is measured.

Cavity pressure transducers located near the gate and at the last point of fill provide the only reliable way to associate a shrinkage excursion with a pressure transient. On a typical production machine with clamp force between 1200 kN and 2500 kN, a peak cavity pressure variation of ±5 bar can propagate to a visible 0.05 % shift on a 60 mm plaque. If the injection stroke sensor does not resolve 0.5 mm, the machine may not detect the shot-to-shot variation responsible for spread. Maintenance conditions including check-ring wear, barrel temperature overshoot, and hot-runner tip temperature differences must be excluded before assigning spread to the PBT compound.

Screw recovery consistency also matters. A cushion variation of 2 mm to 4 mm changes the volume of molten material available for packing, which directly affects shrinkage after ejection because shrinkage is partly a consequence of melt compressibility during the holding phase. Back pressure should be held low and constant, typically 3–8 bar, to prevent gas entrapment without adding excessive fibre breakage. Many cycle-to-cycle investigations begin by recording shot weight because it is a surrogate for packing variation; shot-weight standard deviation below 0.05 % is generally necessary before attributing shrinkage variation to crystallisation differences.

Cycle-to-cycle spread must be reported with the sample size and not as a single largest-to-smallest range. A 10-shot trial can support preliminary mould qualification, but a 30-shot sequence is preferred for process capability because it captures slow drift in hot-runner temperature and screw recovery time. For a nominal 60 mm plaque dimension, a flow-direction standard deviation of 0.03 % equals 0.018 mm. This is small enough to be masked by manual calliper placement and large enough to consume a meaningful portion of a precision-connector tolerance. Statistical reporting should therefore use the same significant figures as the dimensional gauge, and paired flow and cross-flow readings should be kept in the same shot sequence so that process drift is not mistaken for anisotropy.

Table 1. Representative PBT shrinkage ranges measured on edge-gated 60 mm × 60 mm × 2 mm plaques after conditioning at 23 °C and 50 % relative humidity for 24 h.

Table 1 values are not a substitute for tooling data. They represent a process window based on supplier technical literature for standard injection-moulding grades. Published data for a specific multicavity configuration, gate seal time, and fibre length distribution is limited and must be generated on the actual mould because the flow-to-cross-flow ratio shifts with fibre breakage from the compounding extruder and injection screw. Fibre length attrition in a 32:1 L/D reciprocating screw can reduce the anisotropy ratio by lowering the effective reinforcement aspect ratio. This effect cannot be read from a datasheet; it is visible only in paired measurements.

When a 60 mm × 60 mm × 2 mm Plaque Is Measured Without a Fixture, Orientation Bias and Operator Variability Persist

When a 60 mm × 60 mm × 2 mm plaque is measured without a rigid locating fixture, manual placement error can inject 0.02 mm to 0.04 mm of apparent dimension change. For a 60 mm reference length, 0.03 mm is equivalent to 0.05 % shrinkage error, which is comparable to the cycle-to-cycle spread being investigated. A comparator stand with a linear variable differential transformer or digital micrometer having 0.001 mm repeatability, and a fixture that contacts the same edge locations on every cycle, is required. Measurement locations should be marked on the plaque or controlled by an edge gate; measurements taken at different positions along the length can reflect local packing gradients rather than representative shrinkage.

If a vision measuring machine is used, edge detection must be configured for the PBT plaque surface roughness and glass-fibre surface pits. Backlit measurement can produce a different apparent edge than tactile probing because the plaque edge is not a perfect vertical plane and may exhibit draft or flash. A tactile digital comparator with a spherical probe tip diameter of 1 mm to 3 mm reduces surface-defect sensitivity. The shrinkage value is only as reproducible as the edge definition; edges must be measured at positions not affected by gate vestige or ejector-pin witness marks.

Conditioning must be stated. PBT plaques continue to undergo secondary crystallisation after demoulding, and dimensional readings at 1 h and 24 h are not interchangeable. In a production environment, the 24 h value is used for tool correction, while the 1 h value is used for process monitoring. If measurements are taken before the plaque has cooled to 23 °C, thermal expansion inflates the apparent moulding dimension and understates shrinkage. Conditioning under ISO 291, 23 °C and 50 % relative humidity, is the minimum baseline for comparison across trials; conditioning at elevated relative humidity above 60 % can introduce moisture-related dimensional change and should be reported separately.

Table 2. Verification checklist for PBT shrinkage measurement.

Post-moulding dimensional change should be expected. PBT has a secondary crystallisation response to slow cooling, and slow further crystal perfection at ambient temperature is detected as dimensional contraction over 24 h to 72 h. This is why shrinkage measurements reported without elapsed time are not comparable. A measurement protocol should use a fixed time stamp after ejection, not a variable shift-end measurement. For tool corrections, measurements at 24 h are preferable because they correspond more closely to the long-term dimension of the part in service.

Drying is not optional for PBT if cycle-to-cycle spread is to be separated from degradation. Undried PBT containing moisture above 0.02 % hydrolyses in the barrel, reducing molecular weight, lowering melt viscosity, and shifting packing behaviour. The result is a slow drift in both flow and cross-flow shrinkage over a 30-shot trial. The effect can be mistaken for machine drift. A desiccant dryer with a dew point of −30 °C or lower is appropriate for production-scale PBT processing; a hopper hot-air dryer alone may be insufficient when ambient humidity is above 60 %.

Operational boundaries include melt temperature and mould temperature. PBT should not be processed at melt temperatures above 270 °C for extended residence time because thermal degradation changes melt viscosity and packing behaviour. Mould temperatures below 60 °C should be avoided for tight-tolerance parts because rapid cooling freezes low-crystallinity structures that later densify. Hot-runner systems with dead zones also contribute to intermittent black specks and shot-weight fluctuation; this appears in the shrinkage dataset as a non-random outlier rather than a symmetric dispersion.

Shrinkage is not identical to warpage. Warpage arises from a gradient in shrinkage or modulus across the part, not from uniform contraction. Flow-to-cross-flow difference is one driver. A plaque may have a mean flow shrinkage of 1.6 % and cross-flow shrinkage of 2.0 %, yet remain flat if the shrinkage gradient is balanced. The measurement report should state whether flatness is within specification, because a flow and cross-flow shrinkage pair cannot by itself predict whether a PBT moulding will remain dimensionally stable.

In a multicavity PBT mould evaluated on a 1800 kN injection moulding machine, paired flow and cross-flow readings from 10 consecutive shots are collected after each process change. The flow-direction standard deviation is compared with the cross-flow standard deviation; a cross-flow ±3s interval exceeding 0.12 % while flow remains below 0.08 % indicates that the packing profile or cavity-wall temperature balance must be corrected. If both axes exceed the limit, screw recovery and check-ring seating are examined before the PBT grade is changed.