Design optimization of auxetic metamaterial for airless tire using finite element method and machine learning

Abstract Auxetic metamaterials which are characterized by a negative Poisson’s ratio, exhibit unique mechanical properties, expanding or contracting transversely under uniaxial compression or tension. These materials have drawn significant interest in biomedicine, aerospace, impact protection, and in automotives for airless tires which contribute to sustainability goals by improving material efficiency and reducing waste. Auxetic materials are designed using conventional methods such as designs of experiment where microstructures with geometric parameters are linked to specific mechanical properties. However, designing based on desired properties offers greater flexibility. However, it poses challenges due to the potential for multiple microstructure configurations to satisfy a target property. Topology optimization, while considered as an effective design tool, it was found that it is computationally intensive and complex. To overcome these limitations, machine learning (ML) – based optimization offers a promising alternative due to its efficiency and adaptability. This study focuses on the Z-shaped re-entrant honeycomb structures and their design for specific negative Poisson’s ratio, critical for applications such as sustainable airless tires. An artificial neural network (ANN) is trained to establish the relationship between microstructure parameters and Poisson’s ratio. The data set used to train the ML model was prepared by solving large number of finite element models with varying parameters.