# Development of PP Compound Recipes Using Genetic Algorithms and Analytical Models

## TL;DR

This paper uses genetic algorithms and analytical models to design polypropylene compound recipes, successfully replicating key material properties with some limitations.

## Contribution

A novel methodology combining genetic algorithms and analytical models is proposed for optimizing polymer compound formulations under material constraints.

## Key findings

- Analytical models achieved high accuracy for shear viscosity and tensile modulus predictions but struggled with impact strength.
- Genetic algorithms generated three recipes with varying success in replicating target properties under material constraints.
- The methodology is adaptable for including additional optimization criteria like cost or sustainability.

## Abstract

This study explores the development of polypropylene (PP) compound recipes using analytical models (AM) combined with genetic algorithms (GAs). A talcum-filled PP compound, commonly utilised in injection moulding for packaging applications, served as a reference material, with shear viscosity, tensile modulus, and impact strength selected as target properties for replication. The AM were adapted and fitted to a dataset of 52 compounds, achieving high predictive accuracy for shear viscosity and tensile modulus, while impact strength proved more challenging due to its inherent variability. Three recipes were generated using GA under predefined material constraints. Recipe 1 aimed to replicate all three target properties, achieving a balanced compromise with maximum deviations of 13.14% for tensile modulus and 12.37% for impact strength while closely matching shear viscosity (maximum 9.8% deviation). Recipes 2 and 3, focused solely on matching shear viscosity and impact strength, demonstrated exceptional accuracy for shear viscosity, with Recipe 2 achieving near-perfect alignment (2.5% deviation). However, neither recipe approached the tensile modulus target due to material limitations. The findings demonstrate the effectiveness of combining AM with GA for designing alternative formulations, emphasising the importance of realistic targets and material constraints. This methodology is highly adaptable, allowing for the inclusion of additional optimisation criteria such as cost or sustainability. Future work will explore broader material sets and properties, extending the framework’s applicability to technical polymers and diverse industrial applications.