In order to predict fatigue crack propagation at very long life a piezoelec
tric fatigue machine was built in our laboratory. This device is able to fa
il 10(10) cycles and to determine thresholds up to 10(-9) mm/cycle. The : s
pecimen at 10 originality of this machine resides in the possibility to app
ly simultaneously to a specimen a constant tension effect and longitudinal
ultrasonic vibratory:effort with adjustable amplitude. That is to say, the
R ratio can be varied:from -1 to 0.9. It appears that the fatigue threshold
s-are about the-same in conventional fatigue and in resonant fatigue if the
computation of the stress intensity factor K is correct. But there is a ve
ry large difference between the endurance limits at 10(6) cycles and 10(9)
cycles. It-means about 10% for steels and 30% for aluminium and nickel-base
d alloys.
To improve the relation between thresholds and fatigue limits, the gigacycl
e fatigue is studied in a Iii alloy manufactured by powder metallurgy with
two Sizes of inclusions. The testing temperature is 450 degrees C. The firs
t conclusion is that there is:not any infinite fatigue limit until 109 cycl
es. The second conclusion is a possible correlation between the fatigue lim
it and the threshold at R ratio = 0. This relation is much more improbable
for R = -1 because an incubation phenomenon exists around an inclusion or a
porosity loaded at low stress level.