Predicting the influence of weave architecture on the stress relaxation behavior of woven composite using finite element based micromechanics
Authors
Advisors
Issue Date
Type
Keywords
Citation
Abstract
Advanced textile composites employed in automobile and aerospace industries are heterogeneous and viscoelastic in nature. In an attempt to predict its true macro-level viscoelastic response, a finite element based micromechanics model of 5320-8HS (8-harness satin weave) woven composite was assembled to capture the effects of the individual constituents and their microstructure. The study was focused on the influence of weave architecture over its effective response. The model was developed from the microscopy of woven composite cross section using sub-cell modeling approach. It is idealized to contain a linearly viscoelastic matrix and orthogonally interlaced unidirectional composite tows with undulation and floating regions. All the free surfaces are subjected to kinematic conditions of periodic symmetry to recreate Representative Unit cell (RUC) of 8-harness satin weave architecture. Constitutive model of viscoelastic material assumed to follow Prony series. Time-dependent Poisson’s ratio is also incorporated in the unit cell model with an assumption of constant bulk modulus. Experiments are performed on the 5320-1 EO (Extended Out time) epoxy resin and 5320-8HS woven composite under uniaxial tension and 3-point bending. Problems related with axial testing are summarized and results are compared with model predictions. Utilizing the finite element analysis of microstructure, axial and shear stress distribution for both fiber bundle and neat resin was analyzed in the region of contact and maximum stress concentration. Interestingly, the results of current study indicate that shear relaxation of tow undulation and gap region of neat resin along thickness dominate the effective response under tension and shear. Furthermore, stress relaxation predicted by laminate model under 3-point bending for different stacking sequence compare very well with the test data.