AI & Simulation Revolutionising Sustainable Packaging

BMT has developed a cutting-edge approach that integrates AI-driven simulation tools with real-world production processes. This isn’t just a theoretical concept, it’s a proven product and service that derisks production, accelerates time-to-market, and delivers optimised, sustainable packaging solutions. By leveraging these tools, stakeholders across the packaging value chain can achieve lightweight, high-performance designs while reducing material usage, energy consumption, and costly trial-anderror iterations.

The Power of Surrogate Modelling: Predicting Without the Wait

In Stretch Blow Moulding (SBM), simulation transforms what was once a trial-and-error process into a data-driven design tool. Traditional methods waste time and material, relying on physical testing to adjust blowing conditions. In contrast, BMT digitally simulates the bottle blowing process, optimising parameters before any physical testing begins. With results accurate to within 5%, this approach delivers precision where others rely on guesswork. Still, even simulations have their limits—detailed models can be computationally intensive, taking minutes or even hours to run.

BMT overcomes this challenge with AI-driven surrogate modelling, a machine learning approach that delivers near-instant predictions based on prior simulation data. Training algorithms on highfidelity virtual SBM simulations eliminates the need to rerun time-intensive models. This results in rapid insight without sacrificing accuracy (Figure 1). This dramatically accelerates decision-making, shortens development cycles, and provides BMT’s partners with a competitive advantage in fast-moving markets.

For example, in a recent case study, BMT optimised a 20.7 g recycled PET (rPET) preform for a BMT bottle design. By leveraging simulation data, a surrogate model was developed to predict mass distribution, top load, and burst pressure (Figure 1). These predictions were based on inputs such as preform geometry, temperature heating profiles, and blowing process conditions. The model demonstrated high predictive accuracy with R2 scores up to 0.95. This level of precision allows engineers to quickly evaluate design options, streamline decisionmaking, and minimise computational costs.

Perfecting Bottle Design: A Smart Optimisation Strategy

The next phase involves multi-objective optimisation using an evolutionary algorithm to identify the optimal balance between weight reduction and performance. This approach uncovers the tradeoffs between key factors, enabling near-instant selection of designs that offer the best compromise. For instance, in the recent case study, the aim was to reduce the preform weight while maintaining a minimum top load of 150 N and a burst pressure of 1.20 MPa. By analysing the Pareto front, a visual representation of the trade-offs, BMT was able to select designs that met these requirements while achieving significant material savings (Figure 2).

To ensure real-world applicability, the design process incorporates variability. For example, adjusting the temperature heating profile by ±1.5 °C replicates typical fluctuations in production, helping to ensure the final design is robust and reliable. This level of precision is difficult to achieve with traditional methods, which often fail to account for real-world fluctuations in manufacturing conditions. For brand owners and manufacturers, this translates into a robust and reliable optimised design that reduces production risks, cuts costs, and instils greater confidence in the final product.

Real Impact, Real Savings: Lightweighting at Scale

The impact of BMT’s methodology is clear (Table 1). In the case study shown, the optimised preform weighed just 18.1 g, a 13% reduction from the original 20.7 g design. Despite the weight reduction, the preform maintained a top load of 165 N and a burst pressure of 1.21 MPa, meeting all performance requirements.

At scale, this translates into significant benefits. For an annual production rate of 25 million bottles, the optimised design eliminates over 65 tonnes of plastic per year, which is equivalent to five fully loaded double-decker buses. And the savings don’t stop at material usage. The lightweight design also required a 5 °C reduction in peak heating temperature (Figure 3), reducing energy consumption during manufacturing and contributing to a lower carbon footprint—an increasingly important metric in today’s climate-conscious economy.