The article collates guidelines concerning experimental construction of the S-N fatigue curves within a limited fatigue life range. An attempt is made to compare these guidelines, based on experimental data recorded during rotating bending of a notched specimen made of 42CrMo4. The recorded differences in fatigue life values between the constructed curves reach the maximum of 12.2%. According to the above guidelines, the number of tests in particular test series varies from 6 to 28. Based on the performed analysis a conclusion was made that the increase in the number of tests leads to the increase of accuracy but, on the other hand, remarkably increases the time of the experiment and, consequently, its cost. In this context, it is the research worker who, taking into account a possible future use of the fatigue curve, should individually decide about its accuracy.
This paper presents an experimental method for estimating the fatigue limit of trabecular bone using a single trabecular bone sample, the microstructural parameters of which were determined by microCT. Fatigue tests were carried out using the Locati method, with stepwise increasing load amplitude. The fatigue limits of the trabecular structures were determined experimentally in accordance with Miner’s law of fatigue damage accumulation, based on the parameters of the reference S-N curve taken from the literature. On the basis of the fatigue limits, the S-N curves were determined for the tested samples, and from them the compressive strength USS-N corresponding to the fatigue limit for the N = 1 cycle. Ultimate compressive strength US was determined as a result of compression to failure tests. Computational dependencies combining the BV/TV index with US and the BMD index with US were formulated. To verify the proposed method, two groups of human trabecular bone samples were analysed: n = 42 were tested under monotonic loading, and n = 61 were tested under cyclic loading with stepwise increasing amplitude. The statistical test of the distribution conformity of the calculated USS-N compressive strength to the experimental US ultimate strength was performed. The results of the Kolmogorov–Smirnov statistical test were D = 0.19 (p = 0.314). The agreement of the distributions of BV/TV, as determined experimentally and calculated from the computational dependencies, was also tested statistically, with the result of the Kolmogorov–Smirnov test being D = 0.286 (p = 0.065). A similar analysis performed for BMD yielded D = 0.238 (p = 0.185).