PSI - Issue 14

Ilyin A.V. et al. / Procedia Structural Integrity 14 (2019) 964–977 Ilyin A.V., Filin V.Yu. / Structural Integrity Procedia 00 (2018) 000 – 000

973

10

When hydrogen mechanism of corrosion cracking is realized, the more complicated interaction of “mechanical” and “diffusive” factors is observed, and the “most damaging” frequency monotonically reduces with K Imax increase. The description of this process is known from literature but difficult for interpretation within in terms of cyclic strength design. Therefore the simplest and at least sufficiently conservative method to take into account the possible hydrogen cracking is the use of a critical SIF value at static load in a corrosion medium K Iscc for cyclic strength assessment. It is accepted that the stage of a main crack growth is restricted by the condition K Imax > 0.7 K Iscc and fulfillment of the same is considered as the limit state of a structure. It is widely known that an essential difference of critical SIF values in air and in seawater for low-alloyed steels is only typical when their yield stress exceeds 1000 MPa. However for metal of welded joints (HAZ and cast structure of weld metal) a decrease of K Iscc is probable with a lower strength level. Experiments show that the danger of time-based corrosion cracking not related to cyclic loads exists when an anticorrosive protection is not optimal and/or lamination type crack propagation is observed in a structural-anisotropic metal of low purity where hydrogen “traps” exist. For the development of an assessment procedure and its engineering application it is convenient to divide the damage process into two stages:  Stage 1 from fatigue crack nucleation to its propagation to the depth of 2 to 3 mm with a cyclic life defined as N i .  Stage 2 corresponding to the “residual” service life and a crack propagati on from 2 or 3 mm in depth to the ultimate size dangerous in respect of a static strength condition in air (when K Imax approaches K Ic ) or in a corrosion medium ( K Imax approaches K Iscc ). This durability is hereinafter referred to as N c . The suggested subdivision is supported by the following arguments. First, there are different sets of parameters for N i and N c assessment. Second, the suggested definition of N i can be related to a number of cycles corresponding to a visual detection of a crack in a real structure. N i can be taken as an admissible number of cycles for a structure [ N ] under tensile loading. If a structure is loaded by compression stress and/or it has a high structural stress concentration, N c may be significantly higher than N i . However we find that a supposition [ N ] = N i + N c seems insufficiently conservative. The reason is a high probability of occurrence of regions with “unplanned” high RWS level due to deep grinding of welds for their repair or an application of elastic deformation to fit the parts for assembling before welding. In these cases the value of N c may significantly decrease for local parts of a weld. Therefore we accept that in all cases the cyclic life should be determined as     ) /( / ; 0 0 0 0 c i c i N N n n F N N N    . (12) If the major contribution in the total cyclic life is given by N c0 , the mean value N i 0 may be taken for the cyclic life, that is n = 1. When we have a small reserve in residual cyclic life, it is necessary to find the N i estimate corresponding to a low damage probability. Calculations proceeding from the comprehensible fracture probability allow to find the function F in (12) with its maximum value n (0) = 3.0, Figure 6. 4. Unified model of fatigue damage in welded structures and nomograph representation of its procedures

Figure 6 – Suggested dependence of the safety factor vs. a ratio of residual cyclic life tothe total one