Method for evaluating the fatigue strength of welded joints

1. A method for evaluating the fatigue strength of welded joints produced by melting an alloy using a specified welding technique, which method comprises the following steps:

a statistically significant number N of specimens are produced with a welded joint using said welding technique;

each of the specimens is subjected to cyclic fatigue tests, with an alternating stress σ, until failure;

the size of the relevant defect appearing in the fracture plane of the specimens is measured by fractography;

a range of defect dimensions for which the probability of measuring by fractography a defect outside the range is less than a predetermined level of probability t is determined from the measured sizes; and

the lowest stress σ min liable to cause the specimen to fail with a level of probability t is determined from this range.

2. The method as claimed in claim 1 , wherein welded joints are produced on the specimens, the volume of which is between 10% and 100% of the welded joints on one of the parts to be assembled.

3. The method as claimed in claim 1 , applied to evaluating the fatigue behavior of a titanium alloy assembly, the number of specimens of which is between 35and 45, preferably 40.

4. The method as claimed in one of the preceding claims, the level of probability t of which is less than 1%.

5. The method as claimed in one of the preceding claims, the welding technique of which employs a laser beam to melt the alloy.