34 B. Ruellan et al. Fig. 4.2 Description of Signal #1, prescribed in terms of force. The zoom-in indicates that the signal is a sinus Table 4.1 Experimental and calculated fatigue lives for Signals #1, #2, and #3 and for Rbloc Variants #1 and #2 Signal Signal #1 Signal #2 Signal #3 Calculation Ni,calc Variant #1 (withR=0) 97,157 97,157 33,940 Variant #2 (depends on R) 200,332 1,333,081 305,129 Experiment Ni,exp Mean 344,325 1,171,700 574,633 Diabolo 1 410,789 1,221,800 735,769 Diabolo 2 278,029 – 376,389 Diabolo 3 – 977,500 679,959 Diabolo 4 387,309 1,095,900 693,959 Diabolo 5 373,599 1,217,500 552,659 Diabolo 6 306,439 1,140,500 – Diabolo 7 380,929 1,380,800 370,849 Diabolo 8 273,179 1,167,900 612,849 4.5.2 Results The results are given in Table 4.1 and illustrated by the correlation diagram in Fig. 4.3a, b for Variants #1 and #2, respectively. The experimental and calculated fatigue lives are given in ordinate and abscissa, respectively. Individual and mean experimental fatigue lives enable to evaluate prediction accuracy for the three signals. For Variant #1, the fatigue lives are widely underestimated for the three signals. This could be expected provided the fact that for Variant #1 the fatigue lives are calculated from the R=0 endurance curve, regardless of the loading ratio truly experienced by the Diabolo sample. As a result, it does not account for the lifetime reinforcement under nonrelaxing loadings and the fatigue lives are only obtained from the predictor εmax. This is all the more important that the signals investigated are highly nonrelaxing. Note that Signals #1 and #2 have the same predicted lives since they are defined with identical maximum displacements. For Variant #2, the fatigue lives are in fairly good agreement with predictions since they are within the factor two scatter band (see the dashed lines in Fig. 4.3). Typically, the benefits of accounting for the lifetime reinforcement can be illustrated by comparing the predicted fatigue lives between Signals #1 and #2. It is recalled that they have similar maximum loadings, but Signal #2 exhibits higher R. It appears quite clear that Variant #2 is able to account for the lifetime reinforcement, which is all the more important that the loading ratio is high. This result clearly highlights the necessity of accounting for the dependence of Ron the fatigue life in the fatigue lifetime prediction model.
RkJQdWJsaXNoZXIy MTMzNzEzMQ==