PSI - Issue 14

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

970

7

the weld axis. Actually this RWS distribution is already attained after welding of 5 to 7 passes, and for further passes the RWS components were found by Karzov et al (1991) as functions of relative coordinates x/S, y/S . When an asymmetrical groove is used for welding, the epure can essentially change depending on flexibility of fastenings. A type of inherent RWS distribution influences on the kinetics of fatigue process because of its contribution into the total SIF K I for a propagating crack. The function of its component K Ires vs. crack depth a should be found as a result of RWS study. It is assumed in BS 7910 that the above function may be directly obtained from the RWS epures. We have found that the use of residual plastic strain field gives much more expedient RWS assessment. For example, if we calculate K Ires directly from RWS field, for the weld root we get K 1res < 0 due to compression residual stress in that zone, so a fatigue crack must stop at the weld root. On contrary, when this problem is solved by FEM analysis where residual plastic strain field is taken as an initial state, the result obtained by Sadkin et al (2016) corresponds to the experimental observations: a crack extends through the zone of compression RWS making a through-thickness fracture possible. Based on FEM results, the function K Ires ( a ) for multi pass welded joint can be accepted in the kind   a C a S K a C / exp ( ) 2 1 Y Ires      , (8) Coefficients C 1 , C 2 depend on the welded joint type, groove and assessment point. For example, for a crack origin at WTZ in a butt joint with symmetric X-groove the recommended values are C 1 = 1.6, С 2 = 8.1. Local RWS in the area adjacent to the weld, of the width approximately equal to the width of heat affected zone (HAZ), attain a yield stress of material for both stress components if the number of weld passes exceeds 3. Its level can be reduced as a result of overloading, PWHT or surface plastic treatment. The latter method is the most efficient because it forms surface compression stress at WTZ. Reactive RWS can be obtained from FEM results giving “free” shrinkage displacements of joined parts after completion of welding U x 0 and  х 0 . On the basis of these data  R can be predicted as a result of solving an elastic FEM problem for a certain situation. E.g. for a symmetric X-groove butt joint / ] /[ 0 Y 0 R      x p x U S U . (9) A typical case when a high reactive RWS is predicted is welding of reinforcing inserts into a shell structure. In this case an area of tensile stress of 150 to 300 MPa is likely to appear around the weld at the distance up to insert radius plus weld width. Occurrence of reactive RWS as a result of elastic deformation applied to fit the parts for assembling of structures is less predictable. For deep-water structures where the design stress is mainly compressive, the level of tensile RWS may be high enough to change the failure scenario. An appearance of additional RWS that may also be considered reactive one is a probable result of welds repair in the way of deposit after grinding out deep flaws. The full quantitative description of multi-stage fatigue process starting from formation of micro cracks in the volume of a microstructural unit, their shear- mode extension through structural barriers, evolution into “physically short” cracks and further macro cracks is an extremely complicated problem that has not been solved. There fore we have to use the information obtained from two standardized types of test only: ( а) testing of unnotched specimens under a given stress amplitude  a (in a high-cycle area) or a given strain amplitude  a (in a low-cycle area), with a construction of a Wöhler curve, and (b) testing of sharp-notched specimens to construct a cyclic crack resistance diagram, i. e. the dependence of crack growth rate da / dN vs. SIF range  K I . For type (a) tests, the stress-controlled and strain-controlled test results can be combined to construct a common curve σ a c ( N i ). Here σ a c is equal to σ a for high-cycle tests and  a  E for low-cycle tests. Such tests are aimed to find out an effect of the following factors onto the position of this curve:  Actual surface condition, especially for the “as - rolled” state,  Cyclic asymmetry, including the extremely high asymmetry,  Corrosion media and loading frequency. It should be taken into account that the results of these tests include in fact the total number of cycles including 3. Fatigue and corrosion fatigue resistance evaluation of welded joints metal