<p dir="ltr">Defects in composite manufacturing often lead to compromised structural integrity and reduced performance of the final product. A robust constitutive modeling framework is needed to efficiently and accurately predict the deformation responses of dry fabrics and pre-impregnated fibers, paving the way for defect simulation. This thesis presents a comprehensive study on the development and application of a novel constitutive model of fabric preforms and pre-impregnated fibers during composite manufacturing processes.</p><p dir="ltr">This work proposes an integrated constitutive study for textile fabrics in the aspects of mesoscale tow and macroscale fabric behavior. First, a textile architecture-based discrete modeling approach was developed to predict and visualize fiber tow and fabric deformation. The fabrics consist of interlacing virtual fiber tows which are represented by Timoshenko beams joined by translational and rotational springs and rotary dashpots, which are used to capture the energy dissipation during in-plane shear deformation. Second, an anisotropic hyper-viscoelastic model was developed using the strain energy density function of a homogenized unit cell to predict the fabric deformation as a continuous field. A Maxwell model consisting of one Maxwell element and an additional spring is used to consider the nonequilibrium stresses generated during in-plane shear, transverse shear, and through-thickness compaction deformations. Both approaches were experimentally characterized and applied to a hemisphere draping model in the commercial Finite Element Analysis (FEA) software, Abaqus, to demonstrate the predictive capabilities.</p><p dir="ltr">Then, the robust hyper-viscoelastic model is extended to predict prepreg compaction and bending behavior. In the compaction aspect, a coupling term of energy that captures the effect of squeezing flow and a highly nonlinear transverse compression energy are proposed to predict the compaction response of prepreg with liquid and rubbery resin. The viscoelastic parameters were characterized by a Computational Fluid Dynamics (CFD) model for liquid resin and a discrete micromechanics model for rubbery resin. The method was applied to stepwise compaction simulation at different temperatures in Abaqus and compared to experiments for validation. In the bending aspect, the effective shear modulus is expressed as a function of the second-order gradient of deformation. Modeling parameters were characterized by an analytical model that captures the underlying fiber and matrix deformation mechanism. Parametric study was conducted to illustrate the influence of each parameter and the capability to enhance the accuracy of bending prediction.</p>