A Machine Learning Enabled Multiscale Experimental/Computational Framework for Failure of Composite Materials
Abstract
Many applications related to the mission of the Office of Naval Research (ONR) such as hybrid composite structures for both manned and unmanned surface and underwater vessels require modeling of composite materials in the presence of multiple physical phenomena. Reliable and safe design and assessment of systems containing these materials requires advanced multiscale models and simulation tools which can accurately predict behavior and failure subject to external loading conditions. Despite progress made in discovering mechanisms behind dissipative deformations and failure in heterogeneous composite materials, e.g., defects, crack initiation and growth, brittle fracture, debonding at material interfaces and localized deformations, many questions remain unanswered regarding the sequence and interaction of various deformation and failure mechanisms, fiber-matrix interface behavior, and interactions of chemical constituents. In this project, we plan to leverage advances in materials science and artificial intelligence to develop a multiscale framework enhanced by Machine Learning (ML) techniques to reliably capture nonlinear deformations and fracture of composite laminates by linking the atomic and nano-scale mechanisms to larger scales. For this purpose, we will perform multiscale imaging of composite samples using a range of imaging techniques spanning across multiple scales, including Transmission Electron Microscopy, Scanning Electron Microscopy, and X-ray micro-CT to probe the material from atomic to nano- and micro-scales. We will develop a systematic approach for multi-scale characterization and testing of the samples using a variety of techniques such as nano-indentation and micro-indentation tests. Subsequently, we will leverage machine learning techniques to develop a scale-bridging method for linking the atomic scale molecular dynamics simulations to the continuum-scale simulations. The developed multiscale framework will be verified using experimental testresults and will be applied to study the impact of composite constituents and fiber-matrix interface on the failure mechanisms of the material.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 12, 2023
- Source ID
- N000142312180
Entities
People
- Shabnam Jandaghi Semnani
Organizations
- Office of Naval Research
- United States Navy
- University of California, San Diego