A5.: Physically-Informed Neural Networks (PINNs) Phase-Field (PF) Fracture Modelling of Fiber Reinforced Polymer (FRP) Composites

The damage and fracture characteristics of Fiber Reinforced Polymer (FRP) Composites present a unique and intricate challenge distinct from the failure behavior observed in traditional metals [1]. Composite structures exhibit various failure mechanisms during their service life, each with the potential to initiate and propagate independently. In practical scenarios, these mechanisms often act synergistically, manifesting simultaneously [2]. This complexity poses challenges for engineers attempting to comprehend and predict how these diverse failure mechanisms contribute to structural failure. Consequently, engineers’ resort to employing high design safety factors and an increased number of certification tests to account for this uncertainty.

Given the limitations, expense, and time constraints associated with experimental investigations of composites, it is the aim of this project to develop a novel approach for Phase-Field (PF) fracture modeling in FRP composites through the integration of Physically-Informed Neural Networks (PINNs) within the context of Finite Element Methods (FEM). The project aims to enhance the accuracy and efficiency of fracture predictions by leveraging the strengths of both PF methods and Neural Networks (NNs). PINNs will be trained to capture the intricate relationships between material properties, loading conditions, and fracture behavior, enabling a more realistic representation of the complex fracture phenomena in composites.

[1] A. Dean, Phase-Field Modeling of Damage and Fracture in Fiber Reinforced Composites. (Tesis Doctoral Inédita). Universidad de Sevilla, Sevilla, 2020.

[2] A. Dean et al., “A multi phase-field fracture model for long fiber reinforced composites based on the Puck theory of failure,” Compos. Struct., vol. 251, p. 112446, Nov. 2020, doi: 10.1016/j.compstruct.2020.112446.

Supervision: Prof. Dr.-Ing. habil. Raimund Rolfes (LUH), NN (ENS)