NASA NTRS2000
The fatigue crack initiation resistance of an alloy is determined by conducting a series of tests over a range of values of stress amplitude or strain range. The observed number of cycles to failure is plotted against the stress amplitude or strain range to obtain a fatigue curve. The fatigue properties quoted for an alloy are typically the constants used in the equation(s) that describe the fatigue curve. Fatigue lives of interest may be as low as 10(exp 2) or higher than 10(exp 9) cycles. Because of the enormous scatter associated with fatigue, dozens of tests may be needed to confidently establish a fatigue curve, and the cost may run into several thousands of dollars. To further establish the effects on fatigue life of the test temperature, environment, alloy condition, mean stress effects, creep-fatigue effects, thermomechanical cycling, etc. requires an extraordinarily large and usually very costly test matrix. The total effort required to establish the fatigue resistance of an alloy should not be taken lightly. Fatigue crack initiation tests are conducted on relatively small and presumed to be initially crack-free, samples of an alloy that are intended to be representative of the alloy's metallurgical and physical condition. Generally, samples are smooth and have uniformly polished surfaces within the test section. Some may have intentionally machined notches of well-controlled geometry, but the surface at the root of the notch is usually not polished. The purpose of polishing is to attain a reproducible surface finish. This is to eliminate surface finish as an uncontrolled variable. Representative test specimen geometries will be discussed later. Test specimens are cyclically loaded until macroscopically observable cracks initiate and eventually grow to failure. Normally, the fatigue failure life of a specimen is defined as the number of cycles to separation of the specimen into two pieces. Alternative definitions are becoming more common, particularly for low-cycle fatigue testing, wherein some prescribed indication of impending failure due to cracking is adopted. Specific criteria will be described later. As a rule, cracks that develop during testing are not measured nor are the test parameters intentionally altered owing to the presence of cracking.