PSI - Issue 68

Tea Marohnić et al. / Procedia Structural Integrity 68 (2025) 84 – 90 T. Marohnić and R. Basan / Structural Integrity Procedia 00 (2025) 000–000

87

4

Tensile strength R m covered a wide range of values, R m = 845-1723 MPa. Material dataset was further subdivided into two strength subgroups as follows:

• 6 materials with R m < 1000 MPa (designated as low strength, LS), • 17 materials with R m > 1000 MPa (designated as high strength, HS).

Experimental and ANNs-estimated strain-life parameters σ f ′, b , ɛ f ′, c were used to calculate values of fatigue lives 2 N f for 8 different total strain amplitudes ∆ ε /2: 0,25 %, 0,3 %, 0,35 %, 0,4 %, 0,45 %, 0,5 %, 0,9 %, 1,5 %. Fatigue lives pairs with extremely high values of load reversals 2 N f,exp (higher than 5·10 6 ) were not included in the analysis. Usable fatigue lives pairs (2 N f,exp ; 2 N f,est ) were then divided into low–cycle fatigue (LCF) subgroup with 2 N f,exp £ 20000 and high–cycle fatigue (HCF) subgroup with 2 N f,exp > 20000. The same was performed for selected empirical methods. 6. Results Fig. 1 represents experimental vs. estimated number of load reversals 2 N f of low-alloy steels determined with the ANN estimated strain-life parameters for low-alloy steels, while Fig. 2 shows separate diagrams for low-cycle (LCF) and high-cycle (HCF) fatigue lives of low strength (LS) and high strength (HS) low-alloy steels.

Fig. 1. Experimental vs. estimated numbers of load reversals, log(2 N f ) determined with the ANNs for low-alloy steels, all strengths (ALLS) and all fatigue lives (ALLF).

Although all estimates seem reasonable with high percentage of data falling into scatter band of 3 (from 80 % to 93 %), some differences, especially those among trendlines can be observed. If diagrams are viewed together with numerical values of evaluation criteria E f ( s =3), ( E a ) total , ( E a ) Dset and Ē determined for ANNs method, that are presented in Table 1, it can be seen that values of Ē , as an average of previous three criteria are substantially high (over 0,7) when calculated for all materials together and all fatigue lives (ALLS – ALLF), as well as for combinations low strength, low-cycle fatigue (LS – LCF), high strength, low-cycle fatigue (HS – LCF) and high strength, high-cycle fatigue (HS – HCF). However, one subgroup in particular stands out – low strength, high-cycle fatigue (LS – HCF) (upper right diagram in Fig. 2).