A virtual-fiber finite element digital twin for twisted staple yarns

Twisted staple yarns are difficult to predict because short fibers follow helical paths and transfer load through contact and friction. Helical and continuum models capture average twist and orientation effects, but they cannot represent explicit fiber-to-fiber contact, discontinuous load paths, or progressive failure. Fiber-resolved finite element models, often built from micro-CT, can capture these mechanisms, yet they are too costly for large parametric studies. A key gap is a twisted yarn model that is both predictive and computationally practical, and that can quantify how variability at the fiber level controls yarn response under different loading modes. This study presents a virtual-fiber finite element framework that links measured inputs to yarn-scale tensile and bending response, including stiffness, strength, and failure. The resulting model serves as a digital twin of twisted staple yarns, linked to measurements and designed for predictive simulations and systematic updates as new material, friction, or geometry data become available. An adaptive Python modeler in Abaqus/Explicit generates geometry-adaptive fiber centerlines. A truss-and-beam mesh overlay decouples axial and bending stiffness while still capturing fiber bending and realignment in the toe region. Per-fiber property scatter is introduced using dimensionless field variables that scale each fiber’s properties. This enables direct sampling of measured scatter and allows gradual, desynchronized damage. The approach supports uncertainty studies in which yarn-level scatter emerges from fiber-level scatter and contact mechanics. Parametric studies assess mesh refinement, the number of virtual fibers, normal contact stiffness, and friction, together with the distribution and correlation of per-fiber properties. Tensile validation on cotton and wool yarns shows that simulated load–elongation curves remain within experimental scatter and follow the mean response from toe to failure, using both linear-elastic and elasto-plastic fiber constitutive laws. In addition, cantilever bending simulations are compared against experiments using bending rigidity and moment–curvature response, showing consistent trends across twist and material variability. The framework provides a practical properties-to-geometry pipeline and a scalable route to sensitivity and uncertainty analysis for twisted staple yarn design under combined tensile and bending loading, enabling a measurement-linked multi-scale digital twin for industrial textile and fibrous-materials design.