ALE-based finite element modeling of needle-punching using embedded 1D yarns in a 3D preform

Abstract

Needle-punching is widely used to introduce through-thickness reinforcement into fibrous preforms for composite materials. During the process, barbed needles repeatedly penetrate stacked layers and transfer fibers from the in-plane directions toward the thickness direction, progressively creating a three-dimensional reinforcement architecture. Numerical simulation of this process remains challenging because detailed fiber-scale approaches are computationally expensive, whereas continuum-scale models generally do not account for fiber transfer during manufacturing. This work introduces an Arbitrary Lagrangian–Eulerian finite element framework in which transferable fibrous entities are represented by embedded one-dimensional ALE elements, while the surrounding preform is represented by three-dimensional continuum elements. The formulation is independent of the selected scale of the 1D representation: depending on its cross-sectional area and constitutive parameters, a 1D entity may represent an individual fiber, a group of fibers, or an equivalent tow. In the numerical examples presented here, each 1D entity represents an equivalent bundle of fibers potentially engaged by a needle barb. The ALE formulation describes the transport of the represented fibrous material and its internal variables without requiring explicit fiber–fiber contact detection. Elements containing voids and a dedicated update of the 1D connectivity are introduced to represent effective bundle separation, the progressive filling of transfer wells, and successive needle penetrations. For regular needle-punching patterns, the simulations are performed on a representative unit cell combined with periodic boundary conditions. The proposed framework enables the simulation of multiple successive needle punches while accounting for the interaction between the transferred 1D fibrous entities and the surrounding preform. The numerical examples demonstrate that the framework can represent material engagement, through-thickness transfer, localization of damage within the represented 1D entities, partial elastic recovery, retention of transferred material, and the progressive filling of transfer wells during successive needle penetrations. These results assess the capabilities of the numerical formulation but are not presented as quantitative predictions of a specific industrial process. By replacing the explicit treatment of a large and continuously evolving set of fiber–fiber contacts with ALE material transport and an effective interaction law, the proposed framework eliminates contact-search and contact-update operations that constitute a major source of computational cost and numerical difficulty in fully discrete fibrous-network simulations. It therefore provides a computationally tractable framework for investigating the mechanisms governing the evolution of needle-punched preform architectures, prior to application-specific parameter identification and quantitative experimental validation.