Probabilistic characterization for durability assessment under various road strain loads

This study aims to characterize the statistical durability under random strain loads for fatigue reliability prediction of a heavy leaf spring. Characterizing random data involving a probabilistic approach needs to be addressed in terms of defining applicable distribution as the strain data are random cyclic loading that leads to fatigue damage. In this study, random strain loads were extracted repetitively at a sampling rate of 500 Hz for 300 s per block of various road load profiles to obtain the probabilistic features (i.e., the parameters of location and scale, kurtosis, and root-mean-square values). The rainflow cycle counting technique was used to determine the strain-based fatigue life based on the linear damage rule. The load sequence effects computed the highest fatigue life with an estimated range of 1.66×10⁴–3.16×10⁴ cycles/block. The Akaike information criterion proposed that the Gumbel distribution is the most appropriate distribution to be used to model the durability of the leaf spring based on the captured strain signals. From the Gumbel distribution plot, the highest probability of fatigue failure was identified for the highway data in the range of 1.06×10⁶–1.71×10⁷ cycles/block with the probability of 0.72–0.99 since a smooth road surface produced low amplitudes in the strain data. In the reliability assessment, a mean cycle to failure was obtained in the range of 3.07×10⁷–3.57×10⁷ cycles/block, which is the highest value among other data. Hence, the probabilistic approach using the Gumbel probability plot for analyzing fatigue failure under random strain loads provides better statistical characterization in terms of the mean cycle to failure in reliability assessment.