Woven textiles can do more than provide softness and strength. Their microscopic arrangement of yarns can also influence how sound travels through a material. However, finding suitable over-and-under yarn patterns is a difficult design problem. Even a 10-by-10 weave has more possible arrangements than could realistically be tested individually, while a detailed acoustic simulation can take more than an hour.
Now, an international research team has vastly accelerated this process by developing an AI-assisted method to sift through the seemingly endless weave-pattern possibilities. The team comprised researchers from the Laboratory of Tribology and Dynamics of Systems at École Centrale de Lyon/CNRS, the Advanced Institute for Materials Research (WPI-AIMR) at Tohoku University, and RIKEN iTHEMS.
The researchers first generated woven patterns that satisfy basic manufacturing and connectivity requirements. They then grouped the patterns according to four types of repeating symmetry, including rotational and mirror symmetry. An AI system organized the patterns on a design map, placing structurally similar patterns closer together. The team simulated a selected set of designs and used the results to train a prediction model. Once trained, the model could evaluate candidate designs and guide a computer search in less than a minute.
The researchers then evaluated selected designs using independent, detailed simulations. The predicted sound-absorption performance agreed with the simulation results within 5%.
The study also showed how symmetry affects sound absorption. Different symmetries distribute yarn crossings in different ways, changing how air moves through the small gaps between yarns. Some patterns create more winding routes, increasing opportunities for sound energy to be dissipated through friction and heat, while others leave straighter and more open paths. The size of the smallest repeating unit also affects the range of airflow and acoustic behaviors available. This means symmetry can serve as a design parameter rather than simply a visual property.
"By combining symmetry, physics-based simulation, and machine learning, we can explore woven structures that would be practically impossible to examine by trial and error," says lead author Thanasak Wanglomklang. "The method does not replace physics; it uses physics more strategically to identify promising, manufacturable designs."
The team plans to manufacture selected 3D-woven prototypes and test their sound absorption experimentally. They will also extend the method to multilayer woven materials, larger repeating structures and variations arising from manufacturing processes.
This research activity was performed within the framework of ELyT Global, an International Research Network between Tohoku University, INSA Lyon, Centrale Lyon and CNRS. The study was published in Communications Materials on Aug. 27, 2026.
- Publication Details:
Title: AI-assisted symmetry-informed topology optimization of woven materials for broadband sound absorption
Authors: Thanasak Wanglomklang, Frédéric Gillot, Sébastien Besset, and Sonia Mahmoudi
Journal: Communications Materials (Nature Portfolio)
