The objective of this investigation is to predict the residual strength of notched composite Iaminates with various layups, subjected to low frequency fatigue loading with constant amplitude at room temperature, by using a material modeling approach, fracture and fatigue mechanics and the finite element method (FEM).

For simplicity, after thousands of cycles, the geometry of a circular hole of the deformed laminate was categorized as (1) uniformly expanded hole into elliptic shape, (2) crack propagation around the hole transversely. Both types were studied for 12 cases of layups with various proportions of 0, 45, -45 and 90 degree plies. The effect of geometry change during fatigue on residual strength was attributed to the elliptical hole, longitudinal splitting, matrix cracking (reduction moduli of plies), crack propagation and local delamination. Due to the thin through-the-thickness notched laminate, two-dimensional FEM was used and interlaminar stresses were not considered.

Reduction of stress concentration is a reason for the increase of the residual strength of the notched laminate. The stress concentration factor decreases while the elliptic hole becomes more slender; that was examined by the FEM. The residual strength and stiffness were determined by the material modeling with moduli reduction and damaged zone, and the numerical result was obtained by FEM. Laminate theory, point stress criterion, polynomial failure criterion, ply discount method, and fatigue and fracture mechanics (Paris' Power Law) were also included in this research.

Geometry change and moduli reduction are two major effects that are considered to predict the notched strength. The WN point stress fracture model is adopted for simplicity, instead of the average stress criterion. K_tg that corresponds to the unnotched strength in the normalized stress base curve is used to obtain the characteristic length (d_o). We find that K_tg decreases when the elliptic hole becomes more slender and more moduli are reduced (more plies crack). At the time d_o that is determined from K_tg in the base curve is not necessarily a fixed material constant.

The correlation between the fatigue life and the residual strength as predicted by the model and those determined numerically is found within acceptable errors in comparison with the experimental data.