Polybutylene terephthalate reinforced with 30% by weight short glass fibre, commonly designated PBT GF30, warps in injection-moulded form because fibre orientation, anisotropic shrinkage, and gate position govern the direction and magnitude of post-ejection deformation. The cavity-filling flow field aligns fibres along local streamlines, producing a part that does not shrink equally in the flow and transverse directions. In flat rectangular plaques moulded on production-scale hydraulic injection machines with clamp force between 800 kN and 1500 kN, warpage appears as bowing, twisting, or corner lift after the part cools unconstrained in ambient air. The condition is not eliminated by reducing the total mould shrink rate alone; the orientation distribution locked into the frozen skin layer determines how strongly the glass phase restrains contraction of the PBT matrix.

Why Does Fibre Orientation Parallel to Flow Produce Differential Shrinkage in PBT GF30?

Short glass fibres in PBT GF30 have a mean as-moulded length commonly below 400 μm and an E-glass coefficient of linear thermal expansion of approximately 5 × 10⁻⁶ K⁻¹. The PBT matrix, by contrast, exhibits a CLTE of roughly 80 × 10⁻⁶ K⁻¹ to 120 × 10⁻⁶ K⁻¹. During injection, the melt near the cold mould wall solidifies rapidly into a frozen skin layer. In that layer, shear and extensional strain orient the fibres parallel to the local velocity vector. The orientation state is described in process simulation by the second-order fibre orientation tensor, whose principal component a₁₁ can exceed 0.7 in the skin but fall to 0.3 or lower in the core. The 30 wt% glass loading corresponds to roughly 18 vol% fibre when calculated from E-glass density near 2.54 g/cm³ and PBT density near 1.31 g/cm³. Fountain flow at the melt front transports material from the hot core to the cavity surface, and the resulting extensional strain at the front freezes a highly oriented skin before shear can randomise it. This oriented shell acts as a stiff, low-shrinkage layer parallel to flow, while the core contracts more strongly and can drive out-of-plane deformation if the part is not ejected from a fully balanced mould.

Published datasheets for PBT GF30 moulding grades repeatedly place parallel mould shrinkage in the range 0.2% to 0.8% and transverse mould shrinkage in the range 0.4% to 1.2% for plaques of 2 mm nominal thickness. Those values are derived from shrinkage measurements according to ISO 294-4 or ASTM D955. Unfilled PBT typically shows 1.2% to 2.0% in both axes. Table 1 compares commercially reported values. The anisotropic ratio, defined as transverse shrinkage divided by parallel shrinkage, often lies between 1.5 and 2.5 for PBT GF30, but the ratio is strongly dependent on gate type, specimen thickness, and mould temperature. The ISO 294-4 shrinkage plaque is edge-gated with a film gate, so its molecular and fibre orientation is line-like; a production component with a centre pin-point gate cannot be fully characterised by that single shrink value. That limitation is accepted in mould-qualification practice, where flatness is measured separately by coordinate measuring machine after conditioning at 23°C and 50% relative humidity in accordance with ISO 291.

When Gate Position Shifts From Centre to Edge in Flat Rectangular Plaques

Gate position changes the flow pattern that imposes fibre orientation. A centre gate on a flat plaque produces radial or disc-like flow; fibres align along radial streamlines near the gate, but as the front expands, the local flow direction rotates and can leave a biaxially oriented shell with principal shrinkage axes changing across the part. The resulting warpage is usually a symmetrical dome or saddle shape because the radial low-shrinkage direction and tangential higher-shrinkage direction are centred on the gate. An edge gate produces line flow with fibres predominantly aligned along the fill direction; the plaque then shrinks less in the fill direction and more across the fill direction, generating a cylindrical bow around an axis perpendicular to the fill. When the gate is shifted from the centre to a short edge, the distance between the gate and the far side increases, the pressure drop rises, and the far-end corner may experience lower packing pressure. The local solidification rate at the gate freezes the orientation state earlier than at the far edge, so the anisotropic shrinkage is not uniform across the length of the part. In addition, weld lines produced by multiple edge gates introduce discontinuities in fibre orientation and local shrinkage, often called warpage singularities. Production trials with sequential valve-gated hot runners demonstrate that delayed opening of a second gate can reorient the melt front and alter the warpage direction without changing the material grade. Published quantitative warpage data for specific PBT GF30 commercial grades and gate configurations is limited in open literature; moulders typically generate internal gate-location studies using cavity pressure sensors and flatness measurements.

On production-scale injection moulding lines, barrel temperature is normally maintained between 250°C and 270°C for PBT GF30; mould temperature is held between 80°C and 100°C to control crystallinity and post-mould shrinkage. Lower mould temperatures freeze the highly oriented skin more rapidly and can increase differential shrinkage, while higher mould temperatures promote crystallinity and reduce residual stress but may extend cycle time. Pellets are pre-dried at 120°C for 4 h or at 130°C for 3 h in desiccant dryers with a dew point below −29°C; residual moisture above 0.02% causes hydrolysis at melt temperature, lowers molecular weight, and changes the thermal contraction profile. Injection velocity is set to avoid jetting and to control shear rate: high injection velocity increases shear rate in the skin and may raise the orientation tensor principal component a₁₁ in the flow direction. Packing pressure, typically optimised against cavity pressure curves and gate freeze time, compensates for volumetric contraction before the gate seals. Insufficient packing in a thick-walled region leaves under-compacted core material that contracts further after ejection. Cooling time is balanced against ejection temperature; ejection above 80°C can allow more post-ejection crystallisation and greater free deformation. Each of these variables is reflected in production-scale dimensional control charts.

Rheological, Thermal, and Mould Steel Conditions Governing Warpage

Short-glass PBT is a shear-thinning suspension. Published capillary rheometry data place apparent melt viscosity at 260°C and 1000 s⁻¹ in the range 80 Pa·s to 150 Pa·s for standard PBT GF30 moulding grades. The shear rate at the cavity wall during filling can exceed 10 000 s⁻¹ for thin walls below 1.5 mm, strongly orienting fibres near the surface. Because the PBT matrix freezes around 220°C and the glass transition is near 45°C to 60°C, the part passes through a solidification interval in which the fibre orientation state is locked while crystallisation continues. Mould steel selection and cooling layout affect the temperature history. P20 or 1.2738 mould steel with conformal cooling channels reduces the temperature difference between core and cavity surfaces; without such cooling, differential solidification creates asymmetric orientation layers and an additional bending moment. Thermal diffusivity of the mould steel and the interface heat transfer coefficient at the polymer-steel boundary determine the frozen-layer thickness at the moment of gate freeze. In production, mould temperature maps measured by thermal imaging show differences of 10°C to 20°C between gate and edge regions; these differences are sufficient to change local crystallinity and shrinkage.

Commercial injection moulding simulation uses fibre orientation tensors and anisotropic thermal expansion to predict warpage. The Folgar-Tucker model and its reduced-strain closure variants require a fibre interaction coefficient determined by curve fitting polished cross-section data; values for short glass fibre thermoplastics frequently fall between 0.01 and 0.10. Micro-CT analysis of PBT GF30 plaques validates the predicted shell-core orientation gradient and provides local a₁₁ values. The simulation must also incorporate crystallisation-induced shrinkage, not merely thermal shrinkage; for semi-crystalline PBT, the crystallisation shrinkage can be a significant fraction of the total shrinkage. Warpage predictions without measured orientation data and without gate-specific calibration can indicate the correct deformation mode but may underpredict absolute deflection by 20% or more. Published data for specific simulation software and specific PBT GF30 grades is limited; therefore production moulds are validated with physical tooling trials and dimensional inspection rather than simulation alone.

Operational boundaries for warpage control are strict. Melt residence time above 270°C should be kept below 5 min because extended residence time reduces molecular weight and increases post-mould shrinkage scatter. Regrind addition above 25% typically shortens the mean glass fibre length and broadens the shrinkage distribution, increasing part-to-part warpage variation. Chemical exposure to hot alkaline environments, steam, or ethylene glycol at high temperature can attack the PBT matrix and change mechanical restraint; such service conditions may also relax moulded-in stress and release additional warpage over time. Annealing at 150°C for 1 h to 2 h can increase crystallinity and alter flatness, but published warpage data for annealed PBT GF30 at fixed gate locations is limited. In practice, gate position, runner balance, cooling layout, and process settings are evaluated together in short-shot studies and cavity pressure monitoring because no single material property predicts the final distortion in a constrained production part.