V. Syzrantsev, K. Syzrantseva and V. Ilyinykh
Research Journal of Applied Sciences, 2015, 10(8), 334-337.
The study considers developed in kinetic theory of fatigue mathematical model for forecasting of stress-cycled machine parts lifetime with specified probability of non-destruction. The results processing of samples fatigue tests is carrying out on a base of this model. Two parameters of this model: tensile strength and number of cycles before upper inflection point of low-cycle fatigue curve, related with tensile strength by non-linear dependence are random values with unknown distribution laws. In that way left limits of confidence intervals of low-cycle fatigue curve, corresponding to specified probability (1-5%) of samples non-destruction used for forecasting of its lifetime are impossible to calculate. Researchers propose the original algorithm for model parameters determination of fatigue curve passing through any experimental point. Using this algorithm it is possible to calculate during computer modeling the set of fatigue curves. Obtained set of fatigue curves allows generating the random sample the number of cycles before destruction, for any fixed level of actual stress. This random sample is necessary for calculation of its quantile estimation at plotting of limits of confidence intervals. For regeneration of unknown density function of distribution this random value authors use mathematical apparatus of non-parametric statistics ensuring the task decision independently of complexity distribution law of researched random value. The results of developed algorithm realization are illustrated on example of limits of confidence intervals determination for results of low-cycle fatigue test of flexible pipes samples made of HS80 steel.