1972 年 21 巻 227 号 p. 766-771
The random and program loading tests and the tests with constant amplitudes were carried out of U-notched specimens of S50C carbon steel under plane bending stress condition with the mean stress zero. The load programs were so designed that the relative frequency distribution of stresses would be coincident with that of the random load excepting the truncation of higher stresses than 1.5σp, 2.0σp and 2.5σp, where σp was the median of stress peaks. The major conclusions derived from the experiments are as follows;
(1) Even if the higher stress levels appeared in the random loads were incorporated in the load programs, it is difficult to estimate the fatigue strength under the random loads from the results of program loading tests, when the comparison was made on the basis of total number of cycles to fracture. One of the main reasons for this difficulty seems to be caused by the difference of crack propagation rates under respective loads.
(2) The difference in the load sequence such as Lo-Hi-Lo, Lo-Hi and Hi-Lo sequences had no effect on the results of the program loading tests.
(3) Discussions were made on the modified S-N curves which would be laid as the base of life estimations under random and program loads with the assumption of the linear damage rule.