This paper presents a fatigue testing method using a multi-notched test specimen for the purpose of estimating the distribution function of fatigue crack initiation life by a small number of, fatigue tests. Emphasis is placed on the effective use of the fatigue data provided by one test specimen to decrease the number of specimens required in the fatigue tests. The experimental results indicate that two or three multi-notched specimens are practically enough to determine the distribution function of crack initiation life.
Fatigue crack initiation life has a wide scatter and this makes the fatigue design of structural members difficult. In order to make -the fatigue life distribution clear, it is required to prepare a large number of specimens and repeat the fatigue tests under the same loading condition. Such fatigue tests usually take much time and cost. It will take still more time and cost to conduct the fatigue tests such as random fatigue tests or corrosion fatigue tests in long life range. The purpose of this study is to reduce the time(cost) required for the above fatigue tests by applying a multi-notched specimen. The multi-notched specimen is a specimen in which several tens of statistically identical notches are prepared. Heller et al.[l,2] analyzed the life distribution of multi-notched specimen with the aid of extreme value statistics. Itagaki et al. improved the method of experiment and employed Bayes" theorem for the fatigue life analysis. The validity of this testing method is discussed through the fatigue tests using a multicircular- holed plate specimen.
Consider a multi-notched specimen in which " M" number of statistically identical notches are prepared. For all the notches the time to fatigue crack initiation is assumed to follow a distribution function.