The features of the evaluation of the results of experimental data in interlaboratory comparison tests for low-cycle fatigue

Abstract

Improving the efficiency of assessing the results of low-cycle fatigue (LCF) test in the course of interlaboratory comparison tests (ICT) requires a reduction of the uncertainty associated with a number of random factors independent on the quality of testing and qualification of testing laboratories. Those factors are attributed to a random nature of the initiation and development of the fatigue fracture due to the inhomogeneity of the macro- and microstructure of the material, the state of the sample surface, stress concentrator, etc. A method for evaluating the results of LCF tests at ICT is presented. The method is implemented on the basis of linear regression analysis of experimental values of the mean and standard deviation of the logarithm of the durability at a given stress level. The features of evaluating the results of experimental data of interlaboratory comparison tests for low-cycle fatigue are presented. We propose to use the median estimate of the logarithm of fatigue life (corresponding to a given level (amplitude) of stresses obtained from the median fatigue curve constructed from the results of statistical analysis including regression analysis of the variances of experimental data) as an assigned value for assessing the characteristics of functioning of the laboratories participating in the ICT. The assigned value and estimates of the performance of laboratories is obtained using the median fatigue curve, which is plotted depending on the number of participants in the ICT. With a small number of participants, the median low-cycle fatigue curve is plotted according to the test results obtained by the expert laboratory, whereas for a sufficiently large number of participants according to the test results combined for all laboratories participating in the ICT.