Full Article:

Besides advances on failure of composite materials accomplished lately, problems involving damage tolerance are still a field wide open for researches. The state of art theory is still complex and hard to be applied widely in industry level. This work aims to provide a computational tool to calculate composite aeronautical structures such as fuselage and rotor blade considering large damage capabilities. The computational tool is based on the theories proposed by Puck and Matzenmiller, which predict damage propagation under a Progressive Failure approach. It is important to note that even on a mesoscale level, the subroutines separate the mechanisms concerning the fiber and the matrix (inter-fiber) and treat them under different perspectives. Material elastic and ultimate properties are obtained from standardized experimental tests, while for specific parameters, new procedures have been developed. From common tension and compression[0]n,[±45]n,[±67.5]n tests, a package of subroutines analyzes raw experimental data providing these parameters. Additionally, numerical models are calibrated and literature data are adopted. Calculus kernel is coded in FORTRAN (UMAT/URDFIL) and linked to Abaqus®. A high order plane stress material model is employed and different governing laws are applied for fiber and inter-fiber damage. Also, a non-local criterion is used to provide spatial regularization and thus avoid convergence numerical problems. For evaluating consistency at coupon and structural element level, a series of simulations is carried out. Material characterization is checked against experiments on low scale elementary coupons, while implementation is verified by simulating structures in element level. A detailed study of mesh and step size influence is also performed. Finally, validation is held against tests of open-holed and notched plates. It can be seen that the ultimate load is correctly predicted as well as the stiffness during loading. Besides, strain gage response, load cell results and x-ray maps are compared with simulation data. The comparison of results shows that the proposed computational tool can predict the behavior of large damaged composite structures. As future work, a subroutine to analyze delamination can be implemented to improve the numerical predictions performance, for turning this computational tool into a real industry-level virtual testing utility.