Numerical simulation of the microscale impregnation in commingled thermoplastic composite yarns

The impregnation of a glass woven fabric with an amorphous polyethylene terephthalate copolymer (PET-g) matrix was investigated using a finite element (FE) model for interbundle and intrabundle flow of the matrix. Micrographs of samples obtained by film stacking of PET-g to impregnate the glass fabric have confirmed the occurrence of interbundle and intrabundle flow, taking place as separate steps. On the basis of this evidence, two different mechanisms for the fiber impregnation were postulated. The first flow process is associated with a macroscale interbundle impregnation, whereas the second is associated with microscale intrabundle impregnation. Two different FE models were developed to simulate the microscopic and macroscopic flow of the matrix, considering a large number of different random fiber arrangements. Both models could account for the non-Newtonian rheological behavior of the thermoplastic matrix. The microscale impregnation of fibers was simulated by using randomly spaced and nonoverlapping unidirectional filaments. The effect of the number of filaments and the number of random distributions necessary to achieve an adequate accuracy of the method was assessed. The results obtained from the simulation showed that at low pressures, the polymer melt exhibits Newtonian behavior, which makes it possible to predict the tow permeability by the Darcy law. A more difficult situation arises at high pressures because of thenon-Newtonian behavior of the melt. This requires the introduction of a value for the permeability that is also dependent on the rheological properties of the melt. The same non-Newtonian behavior of the matrix was observed for macroscale impregnation of bundles