Advancing Thermoplastic Polyurethane Innovation through Polymerize

The customer aimed to achieve precise targets for 10 specific material properties across two distinct process flows.

They wanted to integrate all their existing data, despite variations in formulation types, whether measured in parts per hundred resin (PHR) or weight percentage (wt%). Additionally, they sought to determine the best approach to effectively incorporate and impart knowledge of the NCO (Isocyanate) index into their predictive models. Another critical objective was to perform predictive analysis on a less-known ingredient by leveraging data points from a well-known ingredient, enabling them to expand their research capabilities and innovate with new materials. The ultimate goal was to develop a robust and flexible modeling framework that could handle these complexities, ensuring accurate predictions and optimized formulations for their diverse production processes.

Key results

• Enhanced accuracy and reduced error: The customer achieved over 95% accuracy and a 5% reduction in average percentage error by integrating advanced modeling techniques and derived parameters, ensuring reliable predictions.
• Flexible prediction approaches: The ability to perform both forward and inverse predictions allowed the customer to adapt their processes by predicting necessary inputs to meet specific property targets.
• Scalability in ingredient predictions: By leveraging Polymerize's models, the customer seamlessly integrated new materials into their production pipeline, maintaining accuracy in predictions for both known and new ingredients.
• Robust model validation: The models demonstrated high predictive power and reliability, consistently achieving a MAPE below 6% and R² values over 0.9, validated through rigorous performance metrics.