Industry Context

Blow-molding is concentrated cost and concentrated opportunity. A modern PET stretch-blow line runs at 60,000–80,000 bottles per hour, with single-line annual output reaching 500 million bottles. Stretch-blow operating parameters — preform temperature, stretch-rod velocity, pre-blow and high-blow pressure, mould temperature — drive wall-thickness distribution, burst pressure and top-load simultaneously, and small operating-point shifts compound at scale. Preform cooling time materially affects section-weight distribution and bottle quality, and isothermal and non-isothermal simulation models predict thickness, stretch-ratio and bottle quality before any physical trial — but few converters operationalise the simulation against the production line.

The commercial backdrop adds pressure. The EU Packaging and Packaging Waste Regulation (PPWR), in force from February 2025, mandates 25–30% recycled-content thresholds in PET beverage bottles by 2030 — and rPET shifts intrinsic viscosity (IV), crystallisation kinetics and the processing window in ways that operator-driven tuning struggles to keep up with. PET resin sits at EUR 1,200–1,500/t and accounts for 60–70% of bottle unit cost; a 5–15% wall-thickness reduction on a 500 M bottle line is a multi-million-euro annual saving before any cycle-time gain is counted.

Problem Statement

• Operator-driven tuning is slow and one-variable-at-a-time — operators adjust pre-blow pressure, stretch-rod timing or oven setting in isolation, missing the coupled interactions that drive wall-thickness distribution.
• Wall-thickness distribution is the master output — it determines bottle weight, top-load capacity (typically 25–35 kgf for a 0.5 L still-water bottle), burst pressure (8–15 bar for CSD) and ESCR resistance simultaneously, and is set in a 2–3 second window during stretch-blow.
• Cycle-time pressure — every 0.1 s of cycle time at 60,000 bph compounds to ~1.7 M extra bottles per year of line capacity.
• Lightweighting initiatives — 0.5 L bottles have come from ~20 g to 8–10 g over the last decade; further gram-shedding requires precision wall-thickness control no operator can hand-tune.
• rPET integration — IV drift of 0.02–0.05 dL/g per recycle pass shifts the optimum operating point; the line needs to compensate continuously, not on a quarterly recipe review.

Why Traditional Approaches Fail

Bench-scale optimization captures only a fraction of the parameter space. Stretch-blow molding simulation in tools like Blowview, ANSYS Polyflow and B-SIM is well-established academically, but integrating it into continuous plant optimization is the operational gap. Operator-tuned lines drift between OEM-recommended set-points and the experienced operator's mental model; new operators take 6–12 months to ramp; and material-lot or ambient-temperature changes trigger another tuning round. Mould surface temperature and residence time materially affect crystallinity and final mechanical performance, but few plants have the closed-loop instrumentation to act on this in real time. The combined effect: most lines run within 10–15% of their theoretical efficiency, with the gap absorbed as scrap, off-spec, or surplus material weight.

The Simreka Solution

1. Process Simulation

Process Simulation builds the blow-molding digital twin — preform geometry, oven profile, stretch-rod dynamics, pre-blow and high-blow pressure curves, mould temperature and cooling time. It runs every operating-point combination in software using thermal-viscoelastic stretch-blow physics, producing a per-recipe prediction of wall-thickness distribution, top-load, burst pressure and cycle time before any physical trial.

2. Hybrid Modelling

Hybrid Modelling calibrates the simulation against the plant historian — preform-resin lot data, ambient temperature, mould-wear state, machine drift. Where physics-based predictions diverge from line outcomes, ML residual models close the gap so the twin tracks the actual line, not the textbook line.

3. Virtual Experiment Platform

The VEP supports any material-side optimisation that the process change implies — rPET inclusion fraction, preform-resin grade swap, IPA modification — predicting IV, crystallisation behaviour and stretch-blow processability for each candidate before it reaches the line.

Step-by-Step

1. Build the blow-molding twin from preform geometry, machine envelope, resin specs and product targets (top-load, burst, weight).
2. Process Simulation runs the operating-point sweep across oven, stretch-rod, pre-blow, high-blow and cooling.
3. Hybrid Modelling fits the twin to plant historian — lot variability, mould wear, ambient response.
4. Virtual Experiment Platform handles material-side optimisation (rPET fraction, IPA, IV grade).
5. Top operating points validated on machine; in-line metrology refines the twin.
6. Continuous learning — every shift sharpens the twin, capturing drift before scrap.

Simreka workflow for virtual blow-molding process tuning — from twin to optimised line.

Simulation Workflow

• Data ingestion: preform CAD, machine envelope (oven, stretch-rod, blow profile), resin specs (IV, IPA, rPET fraction), product targets (top-load, burst pressure, weight), plant historian.
• Model creation: stretch-blow physics with thermal-viscoelastic surrogates per resin grade; Hybrid Modelling residuals against plant-specific drift.
• Iterative optimisation: Pareto across material weight, cycle time, top-load and burst pressure; per-cavity sensitivity surfaced.
• Scenario testing: rPET fraction shifts, IV drift across resin lots, ambient temperature swings, ESCR loading from CSD vs still-water, mould-temperature variation.

Expected Outcomes

• Cycle time reduced 10–20% with retained quality targets — directly raising line capacity at fixed capex.
• Material weight reduced 5–15% via wall-thickness optimisation, with top-load and burst pressure within spec band.
• First-pass top-load and burst-pressure compliance elevated — fewer off-spec pallets, lower QA scrap.
• Continuous re-tuning captures material-lot and machine-wear drift, removing the quarterly tuning round.
• rPET integration handled without manual operator intervention every IV shift.

Blow-molding moves from operator-tuned to engineered optimisation — every set-point traceable to a wall-thickness prediction, every shift's drift caught by the twin before it shows up in QA, and lightweighting and rPET targets met without sacrificing top-load or cycle.

Fill in the form on the left to access the full document.

FAQs

Q1. How does Simreka's Process Simulation differ from a stand-alone Blowview or Polyflow stretch-blow model?

Blowview and Polyflow are physics solvers — Simreka's Process Simulation orchestrates them: multi-variable Pareto sweeps across oven, stretch-rod, pre-blow and cooling; plant-historian fit via Hybrid Modelling; continuous learning per shift. The output is an operating-set-point recommendation tied to OEE, not a single-recipe FEA result.

Q2. Can the twin handle rPET inclusion up to 50% on a 0.5 L still-water bottle without an operator override?

Yes — the Virtual Experiment Platform predicts IV (typical rPET 0.72–0.78 dL/g vs virgin 0.80–0.84), crystallisation rate and stretch-ratio window; Process Simulation re-tunes oven and pre-blow to maintain wall-thickness distribution. The operating-point shift is computed, not hand-tuned.

Q3. Does it support HDPE extrusion-blow containers alongside PET stretch-blow?

Yes — Process Simulation covers extrusion-blow parison drop, programmed wall-thickness profile and clamp-line behaviour for HDPE bottles and containers, alongside PET stretch-blow. The same Hybrid Modelling layer fits the twin to the specific line.

Q4. How is wall-thickness distribution measured into the closed loop?

Where instrumented (capacitive or NIR in-line gauges), measurements feed back into the twin via Hybrid Modelling; otherwise, predictions guide preform and process tuning and weekly offline section-weight sampling sharpens the model. Either way the twin tracks the actual line, not the textbook line.

Q5. How does Hybrid Modelling capture mould wear and IV drift between resin lots?

Hybrid Modelling ingests plant historian — pressure curves, oven temperatures, scrap rates — and fits ML residuals on top of stretch-blow physics. Resin-lot IV shifts and cumulative mould-wear effects are tracked as time-series corrections, so operating set-points adjust before scrap rises.

Q6. What ROI should a converter expect on a 500 M-bottle/year PET line?

An 8% material weight reduction at EUR 1,300/t PET delivers EUR 1.5–3M annual saving on a typical 0.5 L SKU; a 10% cycle-time gain adds 50 M units of line capacity worth materially more depending on contribution margin. The twin pays back inside one quarter on most installations.

Sources

1. SCIRP — Stretch-Blow Molding Parameter Prediction
2. ScienceDirect — Preform Cooling Time Effect
3. Sage — Computer Simulation of Stretch Blow Molding
4. ResearchGate — PET Stretch-Blow Simulation
5. ScienceDirect — Mould Surface Temperature and Residence Time
6. Springer — Multi-response Optimization of Blow Molding
7. European Commission — Packaging and Packaging Waste Regulation (PPWR)