Designing with PET: Shrinkage, Stiffness and Creep Under Load

Polyethylene terephthalate (PET) is a semicrystalline aromatic polyester whose mechanical design envelope is controlled by crystallinity, orientation, moisture and temperature. The glass transition of 75–80 °C and the crystalline melting point of 245–255 °C place unreinforced PET in a narrow processing window for structural components. Shrinkage is not a fixed datasheet value; it ranges from 0.3–0.5% for cold-moulded amorphous parts to above 2.0% in semicrystalline parts. Stiffness follows the same crystallinity dependence, and creep under load becomes the limiting design criterion when service temperatures exceed approximately 60 °C. These variables cannot be treated independently because they share a common origin in the semicrystalline development of the PET chain architecture.

On production-scale injection moulding equipment, the first processing constraint is moisture control. PET pellets are dried in desiccant-wheel dryers with feed-air dew point below -40 °C, at 140–170 °C for 4–6 h, to reduce residual moisture below 50 ppm. A reciprocating-screw injection unit with a low-shear screw and L/D ratio of 18:1–22:1 is typically used, with melt temperature held at 270–285 °C. Residence time above 285 °C is minimised because hydrolytic chain scission, acetaldehyde formation and intrinsic-viscosity loss accelerate sharply. Moulders observe silver streaking, gate blush and brittle parts when moisture exceeds the 50 ppm limit; this is not merely cosmetic but reflects molecular weight degradation. Mould temperature control is therefore the first independent variable that sets shrinkage and residual-stress behaviour.

Why Does Mould Temperature Govern Shrinkage More Than Packing Pressure in PET?

Inside the cavity, PET shrinkage is primarily controlled by crystallisation rather than by packing force. Amorphous PET has a density of approximately 1.33–1.34 g/cm³, while the semicrystalline phase approaches 1.45–1.50 g/cm³; this volume reduction is the direct source of mould shrinkage. When the cavity wall is held at 10–40 °C, the cooling rate is fast enough to suppress spherulitic crystallisation and produces a transparent, highly amorphous part with low, nearly isotropic shrinkage. When the mould wall is held at 120–150 °C, crystallisation proceeds in the cavity, the specific volume decreases, and the part shrinks by 1.2–2.5% depending on section thickness and gate location.

Packing pressure cannot fully compensate this crystallisation-driven volume change because the gate freezes before crystallisation completes. The freezing time is shortened by the same ordered phase that raises viscosity and reduces free volume. Once the gate has frozen, additional screw pressure no longer transfers molten material to the cavity. In multicavity tools, small temperature differences cause cavity-to-cavity shrinkage variation; outer cavities running 10–15 °C cooler than centre cavities can generate dimensional drift of 0.3–0.8%. Shrinkage measurements recorded using ISO 294-4 on 60 mm × 60 mm × 2 mm plaques illustrate this effect.

The glass-fibre data show why filled PET parts bow. A flow-direction shrinkage of 0.2–0.4% adjacent to a cross-flow shrinkage of 0.8–1.2% produces an in-plane differential strain of 0.4–1.0%, which is sufficient to warp sealing faces and change snap-fit engagement. Amorphous PET parts that survive moulding with 0.3–0.5% shrinkage can also undergo post-mould crystallisation if later heated above the glass transition. This secondary shrinkage is not captured by ISO 294-4 unless the specimen is post-annealed. Designers of transparent PET covers that are exposed to hot-fill or dishwasher temperatures must therefore specify either an actively heated semicrystalline process or a post-mould annealing cycle, because the cold-moulded transparent state is dimensionally unstable above 70 °C.

The crystallisation half-time of PET reaches a minimum near 170 °C, but the rate is still slow compared with polypropylene; therefore, cold-moulding at 20–40 °C can fully suppress crystallisation, while hot-moulding at 120–150 °C promotes spherulitic growth. Differential scanning calorimetry of quenched amorphous PET subsequently heated at 10 °C/min typically shows a cold crystallisation exotherm between 130 °C and 150 °C. This exotherm is the clue that amorphous PET parts can shrink and distort if exposed to secondary heating. In multicavity hot-runner tools, channel-to-channel temperature variations of 10–15 °C in the manifold change local fill viscosity and crystallisation onset, producing cavity-to-cavity shrinkage scatter.

Across a 0–45 wt% glass-fibre reinforcement ladder, short-term stiffness and heat deflection temperature shift by roughly an order of magnitude. The values in the following table are representative of injection-moulded plaques tested after conditioning at 23 °C and 50% relative humidity. They are not design allowables for long-term creep, but they establish the initial elastic response.

Tensile modulus is measured on ISO 527-2 type 1A specimens at 50 mm/min or according to ASTM D638-14 at 5 mm/min; the difference in strain rate can shift reported modulus by several percent. Flexural modulus measured by ASTM D790-17 is frequently 10–20% lower than tensile modulus in filled PET because the moulded skin layer and fibre orientation vary through the thickness. Dynamic mechanical analysis at 1 Hz shows that the storage modulus of amorphous PET falls from approximately 2.8 GPa at 23 °C to below 1.0 GPa between 80 °C and 90 °C. This loss of stiffness occurs precisely in the service environment of automotive interior, appliance and lighting components, and it explains why unreinforced PET snap-fits can fail engagement at operating temperature even when they are satisfactory at room temperature.

Fibre-reinforced PET retains a higher fraction of its room-temperature modulus through the glass transition because the glass fibres remain stiff and the crystalline phase acts as physical crosslinks. However, the transverse direction in a fibre-filled part has lower stiffness and higher creep compliance than the flow direction. Orientation generated by filling therefore creates anisotropic mechanical behaviour that must be captured in mould-filling simulation; isotropic material assumptions are invalid for load-bearing designs with fibre loadings above 15 wt%.

At room temperature, moisture absorbed from humid air acts as a plasticiser for PET. At 50% relative humidity, equilibrium moisture uptake of 0.2–0.3% can lower the glass transition by several degrees and increase creep rate. Dimensional change from moisture is smaller than shrinkage from crystallisation, but it becomes significant in precision optical holders and electrical connectors that require dimensional tolerance below ±0.1 mm per 100 mm.

If the Operating Temperature Approaches 70 °C, What Happens to Creep Resistance?

Creep resistance in PET is evaluated under constant tensile stress using ISO 899-1:2017 or ASTM D2990. Short-term tensile tests overstate long-term load capacity because the viscoelastic amorphous phase continues to deform under sustained load. Isochronous stress–strain curves are generated at 1 h, 10 h, 100 h and 1000 h to separate linear viscoelastic response from non-linear creep. For unreinforced PET at 23 °C and 50% relative humidity, the linear viscoelastic envelope generally extends to approximately 20–25% of the short-term ultimate tensile stress. Above this level, creep strain grows disproportionately with stress and the isochronous curves become non-linear.

At 23 °C under a sustained stress of 20 MPa, published tensile creep curves for neat PET indicate that the 1000 h tensile creep modulus falls to approximately 1.5–2.0 GPa, which is 55–70% of the short-term tensile modulus. At 65 °C, the same 20 MPa load can reduce the creep modulus below 0.8 GPa and produce visible part deformation or joint loosening. This temperature sensitivity arises because the service temperature is approaching the glass transition; chain mobility increases, and the amorphous phase cannot sustain load transfer to the crystalline regions. Glass-fibre additions increase the retained creep modulus and reduce total strain, but the cross-flow direction remains the limiting design axis because transverse fibre orientation produces lower modulus and higher creep compliance.

Creep specimen temperature must be controlled within ±2 °C during testing; small deviations near 65–75 °C create large changes in creep rate. In load-bearing enclosures, local stress concentrations at bosses, ribs and knit lines can initiate non-linear creep even when nominal stress is within the linear envelope. Notched Izod values from ASTM D256 are not a substitute for creep data; they are impact indices and do not describe long-term crack initiation under static load. Sharp internal corners adjacent to bolt bosses should be replaced with radii of at least 0.5 times the nominal wall thickness. Weld lines in filled PET are typical creep failure initiation sites, and design stress in these regions is often derated by 25–40% relative to non-weld material.

Time–temperature superposition is often used to accelerate creep prediction for amorphous PET below the glass transition. Horizontal shift factors are fitted from short-term creep tests at 30 °C, 45 °C, 60 °C and 75 °C; a master curve is then constructed for a reference temperature of 23 °C. The method is valid only for the linear viscoelastic region and must not be extrapolated through the glass transition or across cold crystallisation. For fibre-filled PET, creep is anisotropic, so flow-direction and cross-flow-direction specimens must be cut from moulded plaques and tested separately. The anisotropy of creep compliance is typically larger than the anisotropy of short-term modulus because fibre orientation also changes the stress distribution in the amorphous matrix under long-term load.

Published data for neat PET creep under simultaneous hot-water exposure are limited. Hydrolysis attacks the ester linkage, reducing molecular weight and accelerating creep in ways that dry-condition creep curves do not capture. Dishwasher spray arms, hot-well covers and glycol-exposed automotive housings therefore require prototype validation. For such wet environments, a continuous stress below 15 MPa is a conservative starting boundary for unreinforced PET at 60 °C; 30 wt% glass-fibre PET may allow flow-direction stresses near 30 MPa, but cross-flow loading should be evaluated separately.

In a multicavity hot-runner tool for a 30 wt% glass-fibre PET bracket, outer cavities frequently run 10–15 °C cooler than centre cavities when the water manifold is undersized or when hot-runner tip temperatures are uneven. The resulting cavity-to-cavity shrinkage mismatch can exceed 0.4%, causing snap-fit engagement variation and sealing-face warpage. Simulation using Moldflow or Moldex3D with measured PVT data and ISO 294-4 shrinkage values is then used to map the gate-to-fill flow vector, the transverse vector and the local fibre orientation tensor. Corrections usually involve moving the gate to balance flow length, raising steel temperature to 130–140 °C, or increasing glass loading; they do not rely on excessive pack pressure because the gate is frozen before crystallisation ends.

For load-bearing flanges, pump housings and gear casings, the design sequence should use time-dependent secant modulus derived from ISO 899-1 creep curves instead of the short-term tensile modulus. Creep strain at the sealing surface can be limited to 0.5–1.0% over the expected service life by keeping local stress within the linear viscoelastic range and by eliminating sharp transitions. Bolt bosses in unfilled PET should use expansion inserts rather than self-tapping screws when continuous clamping force must be retained for years; thread-forming screws create high hoop stress and are a common creep crack initiation site in amorphous PET housings. For food-contact applications, PET formulations typically comply with FDA 21 CFR 177.1630 and EU Regulation No 10/2011, provided migration testing is completed on the finished part. Structural PET grades containing 30–45 wt% glass fibre are usually outside direct food-contact scope unless a functional barrier is present.

In hot-fill transparent PET containers and overmoulded electronics housings, post-mould shrinkage is deliberately induced by annealing at 130–150 °C for 30–60 min to stabilise dimensions before secondary operations. Unsupported sections can still distort if the annealing fixture does not constrain the part. On production lines, annealing fixtures should be dimensioned to the post-shrinkage drawing, not the as-moulded drawing; otherwise the annealed part will not fit the next assembly station. This is a batch-level design parameter, not a material property, and it must be included in the inspection plan together with ISO 294-4 shrinkage specimens.