Experimental results of subcritical crack growth obtained by a direct
method (the double torsion method) and an indirect method (the dynamic
fatigue method) are compared. Dynamic fatigue tests have been simulat
ed using the results of the double torsion tests. These simulations ex
plain the differences between the results of both kinds of tests and s
how that the results obtained with indirect methods cannot be used to
calculate the lifetime of the pieces. Finally, the simulation is exten
ded to thermal fatigue and the same conclusion is deduced, i.e. the us
ual thermal fatigue tests give an average value of the stress exponent
n which cannot be used to calculate the lifetime of ceramic parts.