PSI - Issue 68

4

CHAHBOUB Yassine/ Structural Integrity Procedia 00 (2025) 000–000

Chahboub Yassine et al. / Procedia Structural Integrity 68 (2025) 310–317

313

Fig. 1. Through wall fracture

In order to economize computation time, we decided to model only a quarter of the mock-up by considering axisymmetry.

a

b

Fig. 2 (a) Modelling of the mock-up FP (b) mesh size near the crack tip

Our FEM model was built with 104,193 nodes and 248,038 elements. The mesh size close to the pre-crack tip was 0.125 mm x 0.0625 mm and was made of eight-node axisymmetric quadratic elements. The most critical part of the model, GTN parameters in the crack propagation area on which FEM simulations have been run, requires combining experimental data with the results given by finite elements. Because of the complexity and high number of elements, each of these simulations took about two days of computation. This required approximately 15 simulations to converge to an optimum set of GTN parameters. For this, the estimated computing time was approximately 30 days. Here, the ferritic mock-up was assumed to respond elastically with a Young's modulus of 203 GPa. In this model, GTN model can make good prediction for the pipeline failure as represented by Figure 3: Force vs. Crack Opening Displacement, and gives good results in terms of crack growth and maximum initiation load. Analysis correspondence of the curve was good at the beginning of ductile tearing that occurred at 2255 kN from simulation, while from the experiment it was 2240 kN. The maximum load predicted was 2741.624 kN at a CMOD of 22 mm, in good agreement with the maximum force experimentally determined of 2808 kN at a CMOD of 23.42 mm. While quite successful, the computation time taken to obtain the right set of GTN parameters took roughly about 30 days.