Origin of lognormal distribution of wear coefficient values

Wear test results exhibit often large scattering and hence poor repeatability, reproducibility, and reliability. The study of spreading of tribological experimental results requires so called big datasets, allowing to understand the origin of scattering and to characterize statistical distributions. In various studies, wear coefficients were found to have lognormal distributions. Therefore, it was supposed that the wear coefficient is affected by two normally distributed variables, which combine through a product, namely the number of contacting asperities and the size of the produced wear particles. In the present study, taking advantage of large sets of measurements collected in the last decades in the Federal Institute for Materials Research and Testing, we demonstrate that a lognormal distribution may arise from a non constant wear coefficient, too, i.e., when a system does not follow Archard's law.

Automated summary

Wear test results often show large scattering, affecting repeatability, reproducibility, and reliability, requiring big datasets for understanding and characterizing statistical distributions. Studies have found wear coefficients to follow lognormal distributions, possibly influenced by two normally distributed variables, and may exhibit lognormal distribution when wear coefficients are not constant.