Every time you drive smoothly from point A to point B, you not only enjoy the comfort of your car, but also the sophisticated engineering that makes it safe and reliable. Beyond their comfort and protective features lies a lesser-known but crucial aspect: the expertly optimized mechanical performance of microstructured materials. These integral but often unrecognized materials are what strengthen your vehicle, ensuring durability and resistance on every trip.
Fortunately, scientists at MIT's Computer Science and artificial intelligence Laboratory (CSAIL) have thought of this for you. A team of researchers overcame traditional trial and error methods to create materials with extraordinary performance through computational design. Their new system integrates physical experiments, physics-based simulations and neural networks to navigate the discrepancies often found between theoretical models and practical results. One of the most surprising results: the discovery of microstructured composites (used in everything from cars to airplanes) that are much stronger and more durable, with an optimal balance between stiffness and toughness.
“The design and manufacturing of composites is fundamental to engineering. Hopefully, the implications of our work extend far beyond the realm of solid mechanics. “Our methodology provides a model for computational design that can be adapted to diverse fields such as polymer chemistry, fluid dynamics, meteorology, and even robotics,” says Beichen Li, an MIT doctoral student in electrical and computer engineering, affiliate to CSAIL, and principal investigator of the project.
An open access paper on the work was published. published in Scientific advances earlier this month.
In the vibrant world of materials science, atoms and molecules are like little architects, constantly collaborating to build the future of everything. Still, each element must find its perfect partner, and in this case, the focus was on finding a balance between two critical material properties: rigidity and toughness. His method involved a large design space of two types of base materials (one hard and brittle, the other soft and ductile) to explore various spatial arrangements and discover optimal microstructures.
A key innovation in their approach was the use of neural networks as surrogate models for simulations, reducing the time and resources required for materials design. “This evolutionary algorithm, accelerated by neural networks, guides our exploration and allows us to find the best performing samples efficiently,” says Li.
Magic microstructures
The research team began their process by creating 3D-printed photopolymers, about the size of a smartphone but thinner, and adding a small notch and triangular cutout to each one. After specialized ultraviolet light treatment, the samples were evaluated using a standard testing machine, the Instron 5984, to perform tensile testing to measure strength and flexibility.
At the same time, the study combined physical tests with sophisticated simulations. Using a high-performance computing framework, the team was able to predict and refine the material's characteristics before it was even created. The greatest feat, they said, was in the nuanced technique of joining different materials on a microscopic scale, a method that involves an intricate pattern of tiny droplets that fuse rigid and flexible substances, striking the right balance between strength and flexibility. The simulations closely matched the physical test results, validating overall effectiveness.
Completing the system was its “Neural Network-Accelerated Multi-Objective Optimization” (NMO) algorithm, to navigate the complex microstructure design landscape, revealing configurations that exhibited near-optimal mechanical attributes. The workflow works as a self-correcting mechanism, continually refining predictions to more closely align with reality.
However, the journey has not been without its challenges. Li highlights the difficulties in maintaining consistency in 3D printing and integrating predictions, simulations, and real-world experiments from neural networks into an efficient process.
As for next steps, the team is focused on making the process more usable and scalable. Li envisions a future where laboratories are fully automated, minimizing human oversight and maximizing efficiency. “Our goal is to see everything, from manufacturing to testing and calculation, automated in an integrated laboratory setup,” concludes Li.
Joining Li on the paper are lead author and MIT professor Wojciech Matusik, as well as Pohang University of Science and technology associate professor Tae-Hyun Oh and MIT CSAIL affiliate Bolei Deng, former postdoc and now assistant professor at Georgia tech; Wan Shou, former postdoc and now assistant professor at the University of Arkansas; Yuanming Hu MS '18 PhD '21; Yiyue Luo MS '20; and Liang Shi, an MIT graduate student in electrical and computer engineering. The group's research was supported, in part, by the Baden Aniline and Soda Factory (BASF).
