Issue 76

M. A. Pascal, Fracture and Structural Integrity, 76 (2026) 49-66; DOI: 10.3221/IGF-ESIS.76.04

In comparison with the linear Model, it shows significantly higher errors with a mean absolute error of 0.1350 mm (3.5 times higher) and a mean relative error of 1.2458% (3.6 times higher), thus the Exponential FNN model demonstrates better performance with lower error rates across all metrics. This result demonstrates that the exponential FNN model accurately tracks the actual degradation path of structural elements. The exponential decay formulation is particularly appropriate for structures undergoing erosion or corrosion processes, where degradation rates typically decrease over time as protective oxide layers form or reactive species are depleted. By aligning with observed degradation trends, this approach provides confidence in the model for long-term prediction, remaining useful life estimation, and determining optimal inspection intervals. F ITNESS - FOR -S ERVICE (FFS) AND STRUCTURAL INTEGRITY ASSESSMENT o evaluate remaining useful life (RUL) across different vessel sections, both linear and exponential degradation models were applied. The linear model assumes constant corrosion rates over time, representing conservative worst case scenarios. The exponential model captures nonlinear degradation behavior influenced by environmental factors and protective mechanisms. Both models were used to predict future thickness and assess structural integrity according to ASME Section VIII and API 579-1 standards [26]. Determine minimum required wall thickness The minimum allowable wall thickness (t min ) for each vessel section was calculated according to ASME Section VIII and API 579-1/ASME FFS-1 [27], accounting for design pressure, material properties, and safety factors. Level 1 acceptance criteria for minimum measured thickness �� − ≥ (0.5 ��� , ��� ) (16) ��� = ( 0.2 ��� ,2.5 ) (17) where t mm is the minimum measured thickness, FCA is the future corrosion allowance (corrosion rate × future service period), t min is the minimum required thickness, and t nom is the nominal thickness [27]. Calculating remaining life = � ������ −� ��� CR (18) where t initial is the initial thickness of the material (mm), t min minimum allowable thickness (mm) determined by fitness-for service criteria, and CR is the corrosion rate (mm/year) [28]. T

Min Predicted Thickness (2040)

Section

t mm

t min

Status

E Head 1 F Head 2 Nozzle A1 Nozzle A2 Nozzle N1 Nozzle N2 Nozzle N3 Nozzle N4

9.1

4.5

6.7

SAFE SAFE SAFE SAFE SAFE SAFE SAFE SAFE SAFE SAFE

59.4

50.6

56.4

7.9 8.5

5.0 5.0 7.0 7.0 7.0 7.0 6.0 8.5

5.8 7.4 9.6 9.3 8.6 9.2

10.8 11.2 11.1 10.8 11.3 15.6

10.3

Shell 1 Shell 2

13.1

Table 4: Exponential model predicted thickness.

60