Monte carlo simulation and modification on the historic probabilistic fatigue S-N curves

Monte Carlo simulation and modification are carried out to rectify the negative prediction of standard deviation of the logarithm of fatigue lives and the absence of confidence assessment of historic probabilistic fatigue S-N curves. An adjustable random numerical generator, mix multiplicative isogenous arithmetical compliment method, is applied in the random simulation process. Sample size for the simulation is controlled in a range of 7 to 15 to address real possible fatigue tests. The negative prediction of standard deviation is corrected to be greater than zero up to the stress level corresponding to a cyclic stressing number of 5. The modification on probabilistic S-N curves is performed by an improved maximum likelihood method. And the curves are reasonably extended to super long life regime. Reasonability and feasibility of the present methodology have been the reconstruction of Q235 steel historic curves. Results show that the historic curves exhibit crossly in high stress amplitude range for the negative prediction of standard deviation, the prediction appears non-conservatively, and the hazard increases with the confidence requirement increasing. After simulation and modification, new probabilistic S-N curves are reasonably extended into super long life regime.