Machine learning for rational design of impact resistant composites
Abstract
Composite materials are widely used in aerospace applications as lightweight structural components due to their superior mechanical properties. With lighter and stronger structural components, the range and mobility of naval aviation aircraft can be increased, ena bling the support of more payload and people during missions. While composite materials have many unique properties, they respond to impact differently when compared to metallic materials. In the scenario of ballistic impact, the damage is localized in the impacte d zone as the impact waves generated outward from the impact point do not have enough time to reach the edges of the entire structur e before the projectile penetrates (or is being stopped by) the structure. Thus, the local behavior of the material near the impacte d zone governs the impact response and due to composites limited ability to undergo plastic deformation, they may fail catastrophic ally. Consequently, the ability to provide ballistic and blast protection is essential in the design of composite armor materials fo r naval aviation aircraft. In this proposal, we aim to design and manufacture next-generation lightweight composite materials with s uperior mechanical properties to provide impact resistance. This project focuses on investigating the fracture mechanisms of novel c omposite architectures beyond the conventional laminate structures to understand fundamental relationships between structure and hig h-velocity impact response. The proposed research integrates multi-physics modeling, artificial intelligence (AI), high-throughput c omputing, and additive manufacturing (AM) to design and manufacture novel composite armors for naval aviation aircraft. With recent rapid advances in AI, machine learning (ML) has been applied to design novel materials. Here, we will not only leverage novel ML tec hniques to design composites with superior mechanical properties but also use ML to automate defect detection and monitoring of AM p rocesses to fabricate reliable and high-quality composite parts. Moreover, we will explore hierarchical design concepts inspired by natural materials such as the mantis shrimp and beetle elytra to improve the ballistic performance of engineering materials. The spe cific aims of this program are as follows: 1) Investigate structure-property relationships of natural materials to reveal the toughe ning mechanisms for high-velocity impact embedded in their hierarchical structures; 2) Develop robust generative inverse design algo rithm integrating physics-based and data-driven approaches; 3) Apply the physics-based inverse design algorithm to composite materia ls and use active learning strategies and reduced-order modeling to accelerate the design process; 4) Manufacture and validate the p roposed composite designs and compare them with conventional laminated composite materials. If successful, the proposed framework ca n lead to the discovery of highly versatile composite materials with superior impact-resistant performance beyond their conventional engineering counterparts, opening up unexplored possibilities in the field of materials science. By integrating advanced computatio nal simulations and advanced ML algorithms, this work is expected to yield significant improvements in composite materials over curr ent state-of-the-art methods to improve safety, reliability, and performance for naval applications.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 20, 2021
- Source ID
- N000142112604
Entities
People
- Grace Gu
Organizations
- Office of Naval Research
- United States Navy
- University of California Regents