- The creation of composite materials has conventionally involved laborious trial-and-error procedures. However, it has always been slow due to differences between theoretical models and actual outcomes.
- The researchers used a high-performance computing framework to run complex simulations in addition to the physical testing, which helped them forecast and improve each composite’s properties before it was built.
A novel artificial intelligence (AI) system has been developed by researchers at the Massachusetts Institute of Technology’s Computer Science and Artificial Intelligence Laboratory to build more resilient microstructured composite materials for use in automobile production.
The project’s Principal Researcher, Ph.D. candidate Beichen Li of MIT CSAIL, stated that the new approach combines physical experiments, physics-based simulations, and neural networks to speed up the creation of new composite materials for application in anything from automobiles to aircraft. Li and his colleagues used the method to create brand-new, very durable microstructured composites that achieve the ideal balance between stiffness and toughness, making them far more resilient than conventional materials.
Li said, “Composite design and fabrication is fundamental to engineering… [and] the implications of our work will hopefully extend far beyond the realm of solid mechanics. Our methodology provides a blueprint for a computational design that can be adapted to diverse fields such as polymer chemistry, fluid dynamics, meteorology, and even robotics.”
The creation of composite materials has conventionally involved laborious trial-and-error procedures. However, it has always been slow due to differences between theoretical models and actual outcomes. The challenge arises from the fact that composites’ atomic and molecular structure determines their strength and endurance. The two most essential qualities that automakers look for, stiffness and durability, must be balanced, as Li demonstrates, and this requires a precise integration of these building components.
The scientists’ methods were described in full in a report published in Science Advances. Their task comprised a broad design space and two types of base materials (one soft and durable, the other hard and brittle), which were combined in various ways to produce the best possible microstructure for a composite.
Li’s team started by creating 3D-printed photopolymers that were significantly smaller than smartphones and had a tiny notch and triangle cut on each one. A typical industrial testing machine was used to assess the photopolymers’ strength and flexibility after exposure to ultraviolet light.
The researchers used a high-performance computing framework to run complex simulations in addition to the physical testing, which helped them forecast and improve each composite’s properties before it was built. Li said his team experimented on a tiny level with the models, combining pliable and stiff materials in precisely the appropriate proportions to produce the perfect balance. They discovered that the simulations were exact, agreeing with the outcomes of the actual tests to confirm their usefulness in practical applications.
Li’s Neural-Network Accelerated Multi-Objective Optimization method was responsible for designing the composite in the first place. Its job was to formulate the molecular structure to have the best mechanical properties. According to Li, the algorithm functions as a “self-correcting mechanism,” generating novel composites, evaluating their characteristics, and then fine-tuning their molecular structure to guarantee that the outcomes of their real trials correspond with the simulations.
“This evolutionary algorithm, accelerated by neural networks, guides our exploration, allowing us to find the best-performing samples efficiently,” Li mentioned.
According to Holger Mueller of Constellation Research Inc., materials design is one of several exciting applications where AI can have a revolutionary effect. “Due to limited simulation options, materials design has traditionally always been a slow and complex process, and it’s going to be interesting to see what the first AI-bonded composites can do for the industry,” he said.
Li acknowledged that his team had significant challenges maintaining consistency between the neural network simulations and the 3D printed composites. He did, however, express optimism for the method he has developed. To apply the approach to materials science in the real world, his team is currently concentrating on making it more scalable and valuable.
“Our goal is to see everything, from fabrication to testing and computation, automated in an integrated lab setup,” Li said.