Research On Fatigue Assessment Method Of Pressure Vessel Welded Joints

Fatigue design is an important issue of pressure vessel design method. Recent years, EN13445-3 and ASME VIII-2 gradually distinguished weld areas and non-weld areas during fatigue design, specialized fatigue assessment methods for welded joint were given in these codes. However, rare related reviews and researches were found in the domestic. At the same time, the traditional Fatigue-Strength-Reduce-Factor (FSRF) method is still used by JB4732 to consider weld fatigue, designers also pay little attention to the welded area during the fatigue assessment.The revision work of JB4732 has been launched, fatigue assessment of welded joints is one of the key issues. We need to discuss whether our country should adopt new method for welded joints fatigue and what method to be used. Thus, it is necessary to systematically study the key technology and safety margin of each method in EN 13445, ASME VIII-2 and JB4732, also the differences in the assessment results between these methods. To support the revision work, this article focused on the theme of welded joints fatigue assessment in pressure vessel and carried out the following basic research work:1) A comprehensive review was made for the ASME Ⅷ-2 FSRF method, EN 13445 Hot-spot Stress method and ASME Ⅷ-2 Master S-N method, based on the code interpretations and related literatures. Key issues related to each method were summarized, including the source of theory, the use of stress parameter, their fatigue curves, thickness correction and safety factors. A more understandable flow chart was given. Then a comprehensive comparison was made from the aspect of fatigue driving force and fatigue resistance. Analysis showed that, all the three methods essentially belong to the category of structural stress method. Specifically, FSRF method is most suitable for engineering application, but theoretical basis of the use of FSRF is questionable. EN 13445 method seems more reasonable with the use of fatigue curves obtained from real welded joints, weld toe failure and weld throat failure are treated differently. The fracture-mechanics-based Master S-N method takes into account the thickness and load mode affects, which is more rigorous in theory.2) The key procedure in the fatigue assessment of welded joints-calculation of stress parameters was studied. The corner joints which show stress singularity were taken as examples, mesh sensitivity of structure stress and primary plus secondary stress results based on different methods were studied. Results suggested that, when applying the stress integration method, the stress average at nodes may lead to smaller structural stress. The elements located at the front side of the weld toe should be selected to calculate the structure stress. Besides, the higher-order reduced integration elements should be selected to reduce the mesh sensitivity. The nodal force method is insensitive to the element type or size, right structural stress result could be reached with even one layer element in the thickness direction. Extrapolation formula has little influence on the hot-spot stress, while element type and size have great influences. Some stress components do not satisfy the equilibrium conditions when stress singularity exists, so the "Effective Linearization Area" should be adopted to obtain primary plus secondary stress. Advices about the stress calculation and modeling were given based on above observations.3) Hundreds of real vessel welded joints fatigue data of several types was collected and reprocessed. Fatigue life prediction accuracy, safety margin of each welded joint fatigue assessment method and the difference between these methods’ assessment results were evaluated by experiment data. Also, attempts were made to discuss several problems which may exist in these methods. Results showed that, FSRF method has sufficient safety margin in the low-cycle region, while the fatigue curve in high-cycle region is not suitable for welded joints. EN 13445 gave the most accurate prediction of fatigue life in this investigation, but the 4% failure possibility suggested that the safety margin is not sufficient for pressure vessels when only take the margin related to the data dispersion into account. Moreover, fatigue limit in EN 13445 is slightly high; The standard deviation of regression in Master S-N curve is relatively small, assessment is not safe. Quantitative analysis of thickness effect suggested that the thickness term may overestimate fatigue strength of welded joints in thin plate.4) Several revision suggestions for the welded joints fatigue assessment method in JB4732 were given..