Predicting strength of fibrous laminates under triaxial loads only upon independently measured constituent properties

• Stress concentration factors in matrix of a composite along three transverse load directions are determined.
• Matrix in situ strengths are defined as originally measured ones divided by the stress concentration factors.
• Composite strengths under 3D loads are predicted based on independently measured constituent properties.
• Favorable correlation between the predicted and measured strengths is obtained.

A long standing and very challenging problem is to predict the ultimate strength of a fibrous laminate under arbitrary load condition only based on the mechanical properties of its constituents measured independently. Although the Bridging Model is unique for calculating the internal stresses in the constituent fiber and resin (which stands for a matrix material throughout this paper) materials subjected to any load including a temperature variation, the in situ mechanical properties of the constituents must be provided beforehand. A unidirectional (UD) composite exhibits a transverse tensile strength smaller than the tensile strength of the monolithic resin material, indicating that the in situ tensile strength of the resin in the transverse direction is different from that measured using monolithic material specimens. This is attributed to a stress concentration. The stress concentration factors (SCFs) of the resin material in a RVE (representative volume element) due to occurrence of the fiber are determined in terms of elasticity theory. The resin in situ tensile, compressive, and shear strengths in the transverse plane are obtained by the corresponding resin strengths measured independently divided by the respective SCFs, whereas the resin in situ longitudinal strengths together with all the other constituent properties are the same as their original counterparts. Using these originally provided constituent properties as input data, the Bridging Model has been applied to analyze the second World-Wide Failure Exercise (WWFE-II) problems. The model's predictions for all the problems have been compared with available experimental data. Favorable correlation has been found.