PSI - Issue 57

Stéphan Courtin et al. / Procedia Structural Integrity 57 (2024) 4–13 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

8

5

• Determination of crack growth rates by Paris law (see Eq. 2):

(2)

where: o a and N are respectively the crack length and the number of cycles o C and n material parameters of the Paris law Note that, as reported by Li et al. (2019), cyclic elastic-plastic stress-strain curves, giving the stress range variation (Δσ) as a function of the strain variation (Δɛ ), are used in the FEA, instead of monotonic stress-strain curves. This point will be detailed later, and sensibility analyses will illustrate the influence of the chosen cyclic elastic-plastic stress-strain curve. In the present study: • Thermal and elastic mechanical parameters are taken from the French RCC-MRx code (2018) and considered at 20°C in a conservative manner. • Different crack growth Paris laws from the French RSE-M code (2014) are tested: mean or conservative, and in air or in Pressurized Water Reactor (PWR) water environment. • Cyclic elastic-plastic stress-strain curves are considered at 20°C, and derived from the French RCC-MRx code (2018) which gives the following relation between the stress range variation (Δσ) and the total strain variation (Δɛ t ) for the considered material (see Eq.3): 3.2. Material parameters where: o and  are respectively Young’s modulus and Poisson coefficient of the material o K and m material parameters given by the French RCC-MRx code (2018) Note that Eq. 3 is debatable because the given relation between the stress range variation (Δσ) and the total strain variation (Δɛ t ) is non-linear and may lead to inconsistencies in terms of Young’s modulus E and ›‹‡Ž† •–”‡‰–Š › ˜ƒŽ—‡•Ǥ That is why a particular focus will be given later, through sensitivity analyses, on the choice of the cyclic elastic-plastic stress-strain curve used in the FEA. 3.3. Crack modelling assumptions In the present study, a series of pre-cracked meshes built with the Zcracks tool are used with a finite element model representing one quarter of the mock-up, considering the conditions of symmetries (see Fig. 4). Two options are investigated later for the crack front shape: a straight one or a free one. The dissymmetry of the experimental initial cracks has not been systematically taken into account and in this case, numerically, the lengths of the initial cracks are the same for both faces of the mock-up. Note also that for the free crack front shape, an optimized strategy is used to choose the finite element nodes driving the crack growth, in order to avoid numerical inconsistencies in building the cracked meshes. (3)