Artificial Intelligence-Assisted Modelling of High-Rate Ductile Fracture

High-rate ductile fracture, particularly in scenarios such as shock loading, poses a significant challenge in engineering, as existing models often fail to represent the complex interplay of plastic deformation, strain localisation, and void formation.

This project seeks to enhance the phase-field method, enabling more accurate predictions of fracture under dynamic conditions. State-of-the-art computational techniques combined with insights from advanced physics will be employed to improve the robustness and applicability of fracture modelling.

Artificial intelligence will support this effort, accelerating parameter calibration and facilitating uncertainty quantification for greater accuracy and reliability.

Submit an informal enquiry Apply for this project HetSys training programme

Supervisors

Primary: Dr Emmanouil Kakouris, Engineering
Prof. James Kermode, Engineering

Project Partner: AWE-NST

A transcript of the video is available by clicking this link - transcript opens in another windowLink opens in a new window

Background

Material failure is a critical area of study in engineering, as understanding how materials behave under stress is essential for designing safe and reliable structures. Dynamic loading conditions, such as impacts or high strain rates, present unique challenges as materials often fail through complex processes such as ductile fracture. In such cases, the material deforms significantly before breaking, involving mechanisms such as strain localisation, where stress concentrates in specific regions, void formation as cavities develop within the material, and crack propagation as these voids link to form larger fractures.

Traditional damage modelling techniques, often struggle to capture these intricate behaviours accurately, particularly in dynamic scenarios. Phase-field modelling offers a powerful alternative, providing a unified framework to simulate fracture processes without the need for explicit crack tracking. Incorporating AI-driven tools further enhances these models by enabling efficient parameter calibration and uncertainty quantification, making simulations more accurate and broadly applicable in engineering contexts.

Project Objectives

The PhD project focuses on developing computational models to better understand material failure under dynamic loading, particularly in ductile fractures. Using phase-field techniques, it will explore processes such as strain localisation, void growth, and crack propagation. AI-driven tools will support efficient parameter calibration and uncertainty quantification, ensuring improved accuracy and practical applications in engineering.

Outcomes

Developed a phase-field model for simulating high-rate ductile fractures under dynamic loading conditions.
Enhanced understanding of the interplay between strain localisation, void formation, and crack propagation.
Developed AI-enhanced tools for efficient parameter calibration and uncertainty quantification.
Continuum mechanics theory.
AI and machine learning fundamentals for scientific applications.
Programming for scientific computing (e.g., Python, C++/Fortran).

Skills that the student will acquire

Continuum mechanics theory.
AI and machine learning fundamentals for scientific applications.
Programming for scientific computing (e.g., Python, C++/Fortran).

To discuss this project further contact: Emmanouil.Kakouris@warwick.ac.ukLink opens in a new window

Please note that due to the nature of our project partner's work, nationality restrictions apply to applications for this project.

If you need guidance on this please email hetsys@warwick.ac.uk