PSI - Issue 68

Hamidreza Rohani Raftar et al. / Procedia Structural Integrity 68 (2025) 1066–1073 Hamidreza Rohani Raftar et al./ Structural Integrity Procedia 00 (2025) 000–000

1068

3

MC%CD1"LL+1%-"#D

MC%C P#CL4%-1A

()+2-1%-"#D

!"#A%C#%D(C)C*+%+)A ! !"#ABBC#D"EFCFGCCH,-C ! ./EM12ABBCFGC-HCPP ! 4P5EA2"C"APSA#D"T#A ! .AB"E28CG#AVTA2M:CFGC-C;Y ! =FPSDM"C"A2BEF2CBSAMEPA2 ! ,C)-C.L+D(C)C*+%+)A ! bAEc?CB"#ABBdC.A2BEcACB"#A28"/C ! =/APEMDcCMFPSFBE"EF2C ! >DBA?CF2C!"D2?D#?C4!.@CABCa

P55L-1C%-"#D"6D6+C%7)+DA+L+1%-"#D%+18#-V7+AD ! !"T?:CFGCAGGAM"CFGCADM/CSD#DPA"A#CF2Cg=h ! gE2?E28CFGCC1A:CGDM"F#BCE2GcTA2ME28Cg=h ! AGGAM"CFGCM#E"EMDcCDccF:CAcAPA2"BCF2Cg=h !"))+LC%-"#DC#CL4A-AD

P55L-1C%-"#D"6D:C18-#+D;+C)#-#

e#ABBT#ACFGC/:?#F8A2CFGCH,aCTSC"FCfHC@eD

e#A?EM"A? MDcTA

gAD"T#AB

!"#ABB

=F##AcD"EF2

4M"TDc MDcTA

!"#DE2

lMm

lDm

l5m

Fig. 1. The procedural framework employed in the current study (a) Experimental data on hydrogen environment embrittlement were obtained from published literature. The factors influencing FCG were gathered, considering constant and variable factors. The variable factors were selected as the input (b) a feature selection method was utilized to assess the importance of factors such as hydrogen pressure and alloy elements in influencing fatigue FCG behavior. Following the feature selection outcomes, inputs were prioritized and specifically employed to train machine learning models (c) The performance of diverse machine learning models was then analyzed and compared. Subsequently, the optimal machine learning model was chosen based on error metrics, and a prediction procedure was completed using experimental data to validate the predictive model

2. Materials and methods 2.1. Data collection

Information on pipeline steel samples with diverse chemical compositions, including yield stress (σ Y ) and tensile strength (σ B ), was gathered from published literature and used as a dataset. During the data collection process, compact tension (CT) specimens were specifically selected to ensure consistency and enable meaningful comparisons. All specimens were prepared according to ASTM standard E647. Information on FCG tests was collected, considering hydrogen gas at various pressures at room temperature (RT), following ASTM E647 standard. All collected data were subjected to a constant load range condition (ΔP-constant test), and ΔK values in FCG tests were computed according to ASTM E647 standard [26]. For all FCG tests, a consistent load ratio (R) of 0.1 was used, and specimens were tested in hydrogen gas at a frequency of 1 Hz. Fig. 2 shows the data collected from the literature [11,13,14,20,21,27,28].