A predictive analytical model for the as-manufactured geometry of 3D woven composites

Predicting the mechanical properties of woven composites through analytical or finite element modeling first requires an accurate geometrical model of the unit cell. This geometry is defined by yarn characteristics such as dimensions, spacing, undulation, and cross-sectional shape. Typically, these parameters are derived from experimental measurements of cured composites, for example, via X-ray CT scanning. This dependency constrains the ability to model the internal architecture and predict properties using only pre-fabrication fabric data. To overcome this limitation, this study proposes an analytical model that predicts the key geometrical parameters directly from preform data. The required inputs are the weaving pattern, fabric areal density, fiber density, yarn linear density, yarn weight distribution, and the target fiber volume fraction. The model is validated against published experimental measurements for various fabrics and composites, with predicted parameters showing good agreement. The comparison between the results yielded to a high degree of accuracy, with the majority of predicted parameters exhibiting a relative error of less than 10%. This model provides designers with a practical tool for rapidly iterating and optimizing composite performance, reducing reliance on costly trial-and-error prototyping.