A CUF-Based Data-Driven Simulation Framework for Static Analysis of Fiber-Reinforced Composite Profile Structures

Reduced-order analysis of thin-walled fiber-reinforced composite profiles remains challenging because discrete macroscopic material data must be incorporated into an equilibrium-compatible structural model and assessed against physical measurements. This study presents a data-driven Carrera Unified Formulation (DD–CUF) framework for the static analysis of three-dimensional composite profile structures. The CUF approximation represents the three-dimensional displacement field through cross-sectional expansion and axial discretization. In the offline stage, macroscopic stress–strain states collected at the integration points of a macroscopic reference model are organized as a discrete material database. During the online solution, the state at each macroscopic integration point is matched with database samples by minimizing a distance functional subject to equilibrium and compatibility constraints. Four cross-section–material configurations are examined, comprising L-shaped and T-shaped profiles made of resin and continuous-carbon-fiber-reinforced material. For each configuration, displacement records at two measurement positions were obtained from independent cantilever loading tests using a laser-based measurement system. The measured responses, finite element method (FEM) results and DD–CUF predictions were compared consistently under the prescribed static loading condition. For the 56predictions associated with each database, the mean relative errors with respect to the experimental measurements were 2.83% and 3.13% for n=5 and n=7, respectively. The discrepancies between the characteristic stresses predicted by DD–CUF and the corresponding three-dimensional finite-element results did not exceed 1.56%. These results show generally comparable response predictions under the two investigated database settings and consistent responses over the adopted axial discretizations for the four profile configurations. The proposed framework connects macroscopic offline material-state organization, CUF-based online matching and comparison with the measured displacement data in an implementable reduced-order analysis procedure.