Coupling Non Isothermal Finite Element Simulation with Real Time Core Preform Infrared Profiling

Gradient

Reheating polyethylene terephthalate preforms for stretch blow moulding produces a steep thermal descent from the outer skin to the core. A standard 28-millimetre beverage preform with a 3.8-millimetre wall exhibits a core-to-surface temperature differential of 12 to 22 degrees Celsius when exiting an infrared oven track. The outside surface absorbs short-wavelength infrared radiation directly, while the inner core depends primarily on conductive transfer through a resin exhibiting low thermal conductivity, roughly 0.15 to 0.24 Watts per metre Kelvin.

Process engineers calculating blow kinematics often input a bulk average temperature across the wall, creating immediate discrepancies between numerical stretch predictions and final bottle wall thickness distributions.

Finite element simulations using isothermal assumptions fail to capture the severe strain localisation that occurs during the stretch rod stroke and high-pressure blow stages. Polyethylene terephthalate strain-hardens exponentially once stretched past its natural draw ratio at temperatures between 95 and 105 degrees Celsius. If the numerical simulation assumes a uniform 102 degrees Celsius throughout the 3.8-millimetre wall, the software projects smooth, uniform thinning across the container profile.

The physical preform behaves differently. The warmer outer layer stretches early under low yield stress, while the colder core layer resists deformation, absorbs the axial load, and triggers premature strain hardening.

A bulk temperature assumption across a four-millimetre preform wall shifts predicted axial stretch force by thirty to forty-five percent against load-cell values.

Direct integration of real-time core infrared profiling into non-isothermal finite element routines addresses this divergence. Core infrared profiling uses pyrometer arrays positioned at multiple spectral bands, paired with numerical inverse thermal reconstruction algorithms, to calculate the through-thickness thermal profile before the transfer gripper places the preform into the blow mould. Feeding these through-thickness nodal temperatures into finite element boundary conditions replaces estimated bulk values with measured reality.

The solver assigns discrete yield limits, hardening moduli, and viscoelastic relaxation times to each shell or solid element layer through the thickness.

The consequence of skipping this coupling appears during tool commissioning. An engineer commissions a 48-cavity blow mould based on an isothermal or unverified thermal model. The simulation predicts a base clearance of 1.8 millimetres and a uniform contact surface at a 0.28-millimetre panel wall thickness.

Press trials produce bottles with heavy base chime accumulation, thin upper labels, and stress-whitening along the support ledge. Correcting this balance requires weeks of oven profiling trials, re-cutting stretch rod stroke cams, or modifying blow cavity base inserts at high expense.

A mold designer reads this thermal gradient as a structural blueprint that dictates steel dimensions. When wall thickness variations exceed six percent around the circumference due to uneven oven heating, the resulting bottle exhibits wall thickness variations greater than twenty-five percent. Resolving this discrepancy at the design stage preserves tool integrity and prevents costly modifications to production molds.

Absorption Characteristics across the Spectral Band

Polyethylene terephthalate absorbs infrared radiation selectively based on wavelength. Near-infrared emitters operating at filament temperatures near 2500 Kelvin radiate energy predominantly between 0.8 and 1.5 micrometres. Within this short-wavelength zone, pristine resin demonstrates high transmissivity, allowing radiant flux to penetrate deep into the substrate before absorption occurs.

Medium-wavelength quartz emitters operating at 1200 Kelvin radiate between 2.0 and 4.0 micrometres, where ester bonds cause significant infrared absorption. This absorption confines thermal deposition to the initial 0.5 millimetres of the outer wall.

Optical pyrometry must bypass surface emission to measure internal preform temperatures accurately. Standard broad-band thermal sensors capture radiant emission dominated by the outer skin. Multi-spectral pyrometer arrays exploit narrow transmission bands.

A sensor filtered to 1.65 micrometres reads flux originating from subsurface layers, while a detector at 3.4 micrometres measures surface skin temperature. Solving the radiative transfer equation across these distinct bands permits reconstruction of the through-thickness thermal profile prior to mould closure.

A primary complication in thermal reconstruction is the temperature dependency of the absorption spectrum itself. As polymer chains expand and enter the rubbery plateau between 85 and 115 degrees Celsius, band broadening occurs around ester carbonyl absorption peaks. A calibration curve derived at room temperature yields errors up to 6.5 degrees Celsius when applied to a preform exiting the oven at production speed.

Production sensors require dynamic calibration matrices adjusted for polymer density and preform skin temperature.

The inverse mathematical reconstruction uses a boundary-value solver that maps detected radiant intensities against known internal emission pathways. Ray-tracing modules account for the curved cylindrical geometry of the preform body. Refraction at the curved boundary redirects rays, creating an optical lensing effect that concentrates internal emission toward the pyrometer axis.

Failure to calculate this geometric distortion leads to an overestimation of core temperatures by up to eight degrees Celsius.

Scattering effects from recycled resin introduce another challenge during production runs. Post-consumer recycled polyethylene terephthalate contains residual pigments, particulate haze, and varying copolymer fractions that scatter light across short wavelengths. This scattering lowers radiant penetration depth, concentrating heat closer to the preform skin.

Without real-time core profiling to adjust for these variations, processors face unpredictable material distribution during blow molding operations.

Mesh

Coupling measured thermal profiles with structural deformation models requires consistent spatial mapping between optical pyrometer outputs and finite element meshes. Commercial non-isothermal blow moulding software uses solid-shell or fully integrated continuum brick elements to model preforms. A standard shell model utilizes five to nine integration points through the thickness.

Assigning a single bulk temperature to an entire element eliminates the measured core-to-surface differential. Modern models map continuous polynomial curves from pyrometer arrays directly onto through-thickness integration points.

Spatial discretization across the preform body must resolve thermal variations along its profile. A typical preform requires thirty to sixty elements along its length, twelve to twenty-four elements through its thickness, and sixty to one hundred twenty elements around its circumference to capture stretch kinematics accurately. Mesh resolution becomes critical at the transition zone beneath the support ledge, where wall thickness changes abruptly from the crystallized, rigid neck finish to the amorphous body.

Under Section 7 of ASTM D1525, heat deflection thresholds provide no guidance for non-isothermal stretch regimes because rubbery-state deformation occurs entirely above the glass transition midpoint.

Interpolation routines project the reconstructed core profile through the structural mesh using localized basis functions. Pyrometer systems measure the preform along its length at multiple axial heights while it rotates before entering the mould. The data arrives as a point cloud representing surface and core values along the preform axis.

Mapping software applies radial basis functions to convert this raw data into an analytical temperature field across the finite element mesh nodes.

The initialization script loads these nodal temperatures into the structural input deck before resolving contact and displacement equations. The temperature field determines local material behaviour through thermodynamic constitutive formulations like the Buckley, G’Sell, or Gorlier hyperelastic models. Every Newton-Raphson iteration step evaluates local stress using the specific temperature assigned to that integration point.

Outer layers deform at lower stress thresholds, while inner layers absorb loads until drawing raises their temperature through internal dissipation.

Element formulation dictates numerical stability during large-strain stretching phases. Reduced-integration eight-node brick elements require active hourglass control to prevent zero-energy modes, while fully integrated bricks suffer from volumetric locking under nearly incompressible plastic deformation. Solid-shell elements with enhanced assumed strain formulations offer a reliable balance.

They handle large radial aspect ratios and support multiple integration points across the wall thickness without locking. When run without these mapped thermal gradients, even high-resolution solid-shell meshes fail to predict the wall thickness variations seen in physical production parts.

Rheology

Polyethylene terephthalate deforms through an interdependent mix of nonlinear hyperelasticity, viscoelasticity, and strain-induced crystallization. In the blow moulding temperature window between 90 and 115 degrees Celsius, polymer chain response depends heavily on thermal conditions. A two-degree drop in local temperature increases the initial yield stress by up to fifteen percent.

At the same time, it lowers the natural draw ratio, accelerating the onset of strain hardening.

The G’Sell-Jonas constitutive relation demonstrates this strong temperature dependency through its explicit thermal softening parameter:

Stress equals material hardness multiplied by the thermal softening factor, the strain hardening exponent, and the strain rate sensitivity index. The thermal factor scales as the exponential of negative beta multiplied by absolute temperature. The hardening parameter beta for typical blow-moulding bottle resins ranges from 0.045 to 0.075 per Kelvin.

An 18-degree drop from the warm outer preform surface to the colder core increases local flow stress by a factor of 2.2 under identical strain rates. The outer surface stretches easily, while the colder core bears the initial load during the stretch cycle.

Strain-induced crystallization introduces another non-isothermal variable to the process. When polymer chains align past their natural draw ratio, crystalline domains form within milliseconds. This crystallization locks the material structure, preventing further deformation.

Because cold resin reaches its natural draw ratio at lower absolute strains, the core layer crystallizes earlier in the cycle than the outer surface. Non-isothermal finite element models calculate this progression by coupling crystallization kinetics directly to local temperature and principal strain tensors.

Dissipated plastic work generates internal heat that counteracts conductive cooling during deformation. High-speed stretching at strain rates between twenty and one hundred reciprocal seconds converts up to ninety percent of plastic work into heat. This self-heating increases local temperatures by five to twelve degrees Celsius within fifty milliseconds.

Non-isothermal simulation engines calculate this temperature rise during each time step. The resulting heat reduces local flow stress, producing thermal softening that balances strain-induced crystallization. Without direct coupling to the baseline thermal profile, simulation models miss this interaction entirely, predicting premature strain lockup where production parts stretch smoothly into mold corners.

Sensor Synchronization and Data Pipelines

Capturing accurate temperature profiles on high-speed blow moulding machinery presents mechanical and electronic challenges. A production machine producing 48,000 bottles per hour moves preforms past inspection points at linear speeds between two and four metres per second. Contact measurement is impossible in this environment.

The sensing system must measure, reconstruct, and map thermal data within the fifty to one hundred millisecond transfer window before mould closure occurs.

Modern production lines deploy high-speed pyrometer arrays connected via deterministic industrial communication networks like EtherCAT or PROFINET IRT. Pyrometers operate with response times below one millisecond, sampling thermal emission at two-millimetre intervals along the preform axis. Position encoders on the oven transfer wheel track rotational and linear coordinates.

This alignment ensures sensor readings match exact spatial locations on the physical preform.

1. Signal Acquisition captures analog radiometric signals from multi-spectral photodiode receivers, converting light intensity to voltage across microsecond sampling intervals.
2. Thermal Profile Inversion processes raw voltage inputs through hardware-accelerated algorithms, translating multi-spectral intensities into discrete surface, core, and inner-skin temperature values.
3. Coordinate Mapping matches thermal data to physical preform dimensions using rotational and vertical position feeds from line encoders.
4. Array Validation verifies readings against minimum temperature limits, identifying misaligned preforms or lamp bank failures before injection into the blow mould.
5. Simulation Data Exchange transfers the structured thermal array to process control systems or finite element engines via low-latency industrial communication protocols.

When communication networks encounter transmission delays, synchronization between the physical preform and its digital model degrades. A twenty-millisecond latency at line speed displaces the thermal profile by eighty millimetres along the preform axis. This misaligns data, placing warm body temperatures on the cold neck finish within the simulation deck.

Real-time control loops require dedicated field-programmable gate arrays or real-time controllers to manage signal processing independently of standard computer operating systems.

Environmental conditions inside the blow moulding cell also affect sensor accuracy. High ambient temperatures, oil mist from hydraulic lines, and vibration from mechanical mould clamps degrade optical calibration over time. Pyrometer housings require positive-pressure air purging through clean, dry instrument air lines.

Optical windows must be made from sapphire or zinc selenide to withstand routine production cleaning without altering transmission properties across critical infrared wavelengths.

Preform rotation during heating presents another measurement complication. If preform rotation speeds do not synchronize with oven conveyor speeds, preforms heat unevenly around their circumference. Pyrometer arrays positioned at a single angle miss these cold spots, recording an incomplete temperature profile.

Advanced inspection systems place three pyrometer heads around the preform circumference at the oven discharge. This arrangement measures radial temperature uniformity before the preform enters the blow clamp.

Validation

Validating non-isothermal simulation results against physical bottle production requires methodical, destructive part testing. Simulation models calculate thickness profiles, wall distributions, and volume capacities based on initial temperature conditions. Physical bottles produced from matching thermal runs must undergo dimensional inspection to verify these numerical predictions.

Verification workflows map predicted bottle profiles against physical parts using coordinate measuring machines, non-contact optical scanners, or magnetic Hall-effect thickness gauges. Dimensional measurements are taken across twenty to forty axial points on the bottle sidewall. Discrepancies between predicted and measured wall thicknesses highlight errors in the initial temperature model, material parameters, or stretch rod timing curves.

A three-millimetre vertical displacement between the predicted and observed natural stretch transition indicates inaccurate core temperature assumptions within the structural simulation deck.

Destructive top-load and pressure-burst testing provides further insight into process accuracy. A bottle with correct nominal wall dimensions may fail performance testing if local material drawing occurred at incorrect temperatures. Cold-drawn resin exhibits high strain hardening with low residual elongation, resulting in brittle failure under load.

Warm-drawn resin stretches without developing crystalline alignment, lowering sidewall rigidity and top-load capacity.

Process engineers use these mechanical performance metrics to calibrate non-isothermal simulation models. If simulation predicts acceptable top-load performance while production bottles fail axial crush testing, the error typically traces back to core temperature assumptions. Colder preform cores increase material stiffness along the inner bottle wall.

Non-isothermal simulation decks must account for these through-thickness thermal gradients to predict actual part performance on the production floor.

Can Finite Element Feeds Govern Closed Loop Oven Profiles?

Connecting verified simulation models directly to oven control networks enables automated, closed-loop thermal regulation on production blow moulding lines. Modern stretch blow moulders run under open-loop control, relying on operators to adjust lamp settings when ambient temperatures shift or resin lots change. Closed-loop control systems use pyrometer data, simulation predictions, and downstream bottle inspection to balance oven settings automatically.

Downstream inspection systems scan finished containers using non-contact infrared or optical sensors, measuring wall thickness distributions across the bottle profile. When label panels thin or base chime dimensions drift outside statistical process limits, the control platform identifies which preform zone caused the variation. Rather than relying on simple proportional adjustments, the system runs an inverse simulation to calculate the through-thickness thermal profile needed to restore correct material distribution.

• Lamp Power Modulation adjusts voltage across specific infrared zones, altering radiant output without requiring line speed changes.
• Ventilation Flow Compensation regulates airflow through the oven cavity, balancing convective surface cooling against infrared penetration.
• Preform Rotation Speed Control adjusts spin rates through heating zones to eliminate circumferential temperature variations.
• Stretch Rod Speed Profiling alters servo-driven stretch rod speed curves, matching mechanical motion to the preform thermal state.
• Secondary Blow Timing Shifts adjusts high-pressure blow triggers to accommodate resin temperature drifts without causing structural blowouts.

Implementing closed-loop thermal control alters tooling economics and process qualification workflows. Traditional mold commissioning requires iterative steel modifications to balance wall thickness across the container body. An automated thermal control system shifts this balance to the heating process, allowing standard mold designs to run broader process windows without physical tool adjustments.

This closed-loop integration provides significant advantages when processing post-consumer recycled resins. Recycled material lots exhibit variable intrinsic viscosities and infrared absorption rates, causing continuous thermal shifts inside the oven track. Real-time pyrometer profiling detects these variations at the oven exit, allowing the control system to adjust lamp output before material reaches the blow mould.

This dynamic control maintains container wall thickness uniformity despite variations in raw material quality.

A supplier will often attribute wall thickness variations to resin inconsistency rather than adjusting oven thermal settings. Closed-loop systems eliminate this ambiguity by continuously balancing thermal inputs against measured container dimensions. This automated control maintains process capability across shifts, cuts changeover scrap rates, and reduces cycle-time drift during long production runs.