疲労寿命の統計的分布とこれに及ぼす熱処理の影響

Abstract

In general, the results of the fatigue tests on the metallic material are found to be widely scattered, accordingly, even if a test were performed under the same condition, the values of the fatigue lives obtained would be widely dispersed because of its structure-sensitive properties. Since we can not equalize these experimental factors, a statistical interpretation of the fatigue data would be necessary.In the present study, the authors have carried out the fatigue tests of the low carbon steel (0.23%C) with large sample (many test specimens per single stress level), and from the experimental data, the following three items have been discussed; (i) the effect of the cutting position of specimens taken from the original bar stocks on the fatigue life, (ii) the statistical distribution of the fatigue life under the higher and lower stress levels, and (iii) the effectiveness of the heat-treatments (normalizing and annealing) on the dispersion of the fatigue life, by means of F-test for the fatigue data.As a result, the following conclusion has been obtained:(1) The correlation coefficient (absolute value) between the cutting position of the specimen and the fatigue life amounts to 0.1-0.3 in cases of the as-rolled material and the normalized material, and 0.04-0.06 for the annealed material. It is concluded that the cutting position of the specimen from the original rolled bar has no influence on the fatigue life.(2) Under the higher applied stress level, the statistical distribution of the fatigue life is normal, but when the applied stress is lower, the log-normal distribution, or the extreme value distribution is applicable to the fatigue life.(3) No significant difference of variances of the fatigue life is recognized between the asrolled material and the normalized material, while the variance of the fatigue life of the annealed material is smaller than that of the as-rolled material, under the same applied stress level.