Machine Learning Meets Fracture Mechanics: What Can We Learn?

Fracture mechanics, as an essential field of solid mechanics, has played a significant role in characterising and understanding cracking and failure processes of natural and engineered materials. With the emergence of the 4th paradigm: data-driven/physics-informed methods, machine learning (ML) offers transformative opportunities for this field. With the exponential growth of research leveraging ML in the fracture mechanics community, the need to ‘learn’ complex physics or to map from datasets becomes clearer. To this end, this paper presents a screenshot investigation at the intersection between ML and fracture mechanics. Firstly, the key ML algorithms are scrutinised, followed by the commonly used computational methods. Furthermore, we provide an in-depth review of ML applications across six significant aspects of fracture mechanics. Beyond these core findings, this work identifies underexplored research directions and offers actionable recommendations. This timely review highlights the potential of ML methods to advance fracture mechanics. It provides a practical guide for researchers to incorporate ML techniques into their studies, paving the way for the 4th paradigm in the fracture mechanics community.