It is well noticed in Figure 1 that the values of the confidence index are greater than 0.8, reflecting the good correlation of the experimental data as well as the reasonable adjustment of the Weibull distribution. In addition, the obtained values of β are greater than (1) which shows that the damage due to cyclic loading was distributed over the entire part subjected to cyclic fatigue. On the other hand, the analysis of the reliability of the fatigue data of the non-immersed composite material presents form factor values which gradually increase with the decrease in the fatigue loading ratio. These results show that the fatigue data of composite are scattered for high loading levels, and the dispersion of these data has undergone a remarkable decrease with the decrease in loading level and consequently a symmetric distribution of the fatigue data. Indeed, the application of a low fatigue load causes damage to the material with a large number of cycles, which leads to the conclusion that the slopes of the regression lines effectively reflect the dispersion of the fatigue data for the materials examined.
Using equations (1) and (2) allows the calculation of the [MTTF] and the [SD] as well as the [CV] of all the composite materials tested. These statistical parameters make it possible to evaluate the relative dispersion of the fatigue data for different levels of cyclic loading. Figure 2 shows the effect of [MTTF] on the [CV]. It should be noted in this figure that the fatigue life values are more dispersed for the composite sample immersed in water for a period of 20 days due to the heterogeneous nature of the material and the presence of water in the carbon/polyester composite. However, the widest initial dispersion was recorded over the life range of 101 to 102 cycles for all materials tested.