Speaker
Description
Technical textiles can be described as complex porous media whose performance is governed by coupled air, moisture and heat transport mechanisms across multiple length scales. These transport properties play a critical role in determining thermal comfort, functional efficiency and, in specific applications, user safety. However, their experimental characterisation remains challenging due to the strong dependence on material architecture, fibre arrangement and environmental conditions.
In this context, predictive modelling approaches are increasingly required to support the design and optimisation of textile systems, reducing reliance on time-consuming and application-specific experimental campaigns. Computational Fluid Dynamics (CFD) enables detailed resolution of flow and transport phenomena within textile structures, but its applicability at the product-design stage is often limited by computational cost and geometric complexity. Conversely, Machine Learning (ML) techniques offer fast property prediction once trained, yet strongly depend on the availability and quality of representative datasets.
Hybrid CFD–ML frameworks therefore represent a promising strategy to combine physics-based understanding with data-driven efficiency, enabling accurate and scalable prediction of air permeability, moisture management and heat transfer properties in technical textiles.
In this work, a previously validated workflow for the prediction and assessment of air permeability in technical textiles is extended towards the coupled evaluation of moisture management and heat transfer properties. The proposed framework considers a wide range of synthetic textile geometries, systematically generated by varying key structural parameters such as yarn density, weave pattern, material composition and yarn flattening behaviour.
Textile geometries are generated using the open-source software TexGen, specifically developed for the parametric modelling of textile architectures and the export of STL representations. These geometries are subsequently imported into the CFD solver OpenFOAM, where numerical simulations are performed to resolve airflow, moisture transport and heat transfer phenomena according to the targeted transport property.
While CFD simulations provide detailed insight into transport mechanisms within textile porous structures, their computational cost makes them unsuitable for extensive parametric studies or real-time design optimisation. To overcome this limitation, the CFD-generated dataset is employed to train and validate a Machine Learning model capable of predicting air permeability, moisture management and thermal transport indicators directly from a set of geometrical descriptors.
The resulting hybrid CFD–ML framework combines physical interpretability with computational efficiency, enabling fast and scalable prediction of transport properties in technical textiles and supporting performance-driven material design.
Acknowledgment
This study was carried out within the MICS (Made in Italy—Circular and Sustainable) Extended Partnership and received funding from the European Union Next-GenerationEU (PIANO NAZIONALE DI RIPRESA E RESILIENZA (PNRR)—MISSIONE 4 COMPONENTE 2, INVESTIMENTO 1.3—D.D. 1551.11-10-2022, PE00000004). This manuscript reflects only the authors’ views and opinions; neither the European Union nor the European Commission can be considered responsible for them.
| Country | Italy |
|---|---|
| Acceptance of the Terms & Conditions | Click here to agree |
