Abstract

Low-cycle fatigue failure in titanium metal matrix composites is caused by two separate damage mechanisms: fatigue crack growth in the Ti matrix and fiber breakage. Here, a coupled numerical model for predicting both crack growth and fiber breakage is developed and applied to predict low-cycle fatigue lives in a SiC-fiber reinforced Ti matrix composite. A three-dimensional finite element model containing a matrix crack, nucleated on the first loading cycle in the reaction layer around a fiber, that is bridged by SiC fibers is used to calculate both the matrix crack tip stress intensity factor and the local fiber stress concentrations due to the matrix crack, as a function of the crack size. The crack tip stress intensity factor is used in a Paris-law model for the growth rate of the matrix crack. The local stress distributions in the fibers are used as the effective "applied" load within a three-dimensional Greens Function method that simulates the fiber damage process at any fixed fatigue crack size. Fiber failure preferentially occurs within the matrix crack region, where the fiber stresses are comparatively high, and composite failure occurs when the damage in this region is sufficient to drive fiber failure throughout the remainder of the composite in a crack-like fracture mode. A fatigue life threshold is predicted at about 80% of the quasistatic tensile strength, where the fiber bundle can survive even with a matrix crack extending throughout the entire cross-section. Predictions for the low-cycle fatigue of Ti-matrix (IMI834) reinforced with SCS-6 SiC fibers compare well with available experimental data at high stresses using pristine fiber strengths and no adjustable parameters. Using literature values for the fatigued fiber strength beyond 10(4) cycles and no adjustable parameters, the experimental data are also well matched at lower stresses. The model demonstrates that fatigue life can be dependent on actual composite size and can be very sensitive to initial fiber damage. (C) 2001 Published by Elsevier Science Ltd on behalf of Acta Materialia Inc.

Details