microscopes, ekctrrmic flash light sources, cl~~y operated 35mm cameras and pro~mrn~ components to allow exposures to be mfdc at lnaximum tension load at predetarmined cyclic times without interrupting the fatigue test. Prior to testing, the notch roots of the specimens were polished in a polishing iixture using alumiwm oxide Polish compound. As a result of the polishing process, the. individual 8 Cm diameter graphite fibers were clearly visible and matrix voids and llber dcbonding could easily be seen.
Anomalous fatigue behavior was observed in the limited number of tests which were performed. In two tests where the cyclic loads were approx. 75 and 90% respectively of the static fracture load there were differences in the crack initiation behavior. In the specimen which was tested at 75% of the static load, a macrocrack developed in one of the 600 layers during the lirst cyck of kading yet the specimen was then able to sustain 10,OOOcycles withoutfracture.Inthespecimenwhkhwastestedat9O%of thcstaticloadthcobservations indicattdprogressivcdamageduringtUletestwhichinckdcdfibercrackinginthe00layc~,6~rdebondinginthc 60" layers, and delamination between a pair of adjacent 60" layers. This specimen also sustained 10,000 cycles of cyclic loading. Both specimens were then monotonically loaded to failure and their residual strengths were approximately the same as that of a virgin static specimen.
A third specimen, which WC had planned to cyck at !M% of the static fracture toad failed on the first cyck of the test. A macrocrack developed at 55% of the static fracture load and catastropic failure occured at 7g%.
Possible causes for the apparent m behavior have not yet been isolated. However, there are indications that rnan~ac~~ irregularitks in the material may contribute to the strange fatigue behavior.