Influence of Residual Stress on Fatigue Failure of Welded Joints

Show full item record

Title: Influence of Residual Stress on Fatigue Failure of Welded Joints
Author: Lu, Xiangyang
Advisors: Dr.Tasnim Hassan, Committee Chair
Dr.Vernon Matzen, Committee Member
Dr.James Nau, Committee Member
Dr.Murthy Guddati, Committee Member
Abstract: This dissertation makes an effort to understand the influence of residual stresses on the fatigue failure of butt- and socket-welded piping joints. In order to achieve this goal, this study develops experimental fatigue failure data and a numerical scheme for the analysis of these data. A set of low-cycle fatigue tests of welded piping joints in the cantilever setup is conducted. A new observation made in these tests is that the recorded strains near the weld toe ratchet continuously, which results in the initiation of fatigue crack(s). Comparison of these ratcheting responses with those from the cyclic pipe bending and material ratcheting experiments indicates that the residual stresses at the welded joints may not relax to zero after a few inelastic cycles, as assumed in the fatigue design methods. This observation is further supported by the symmetric strain response (no ratcheting) at the mid-pipe length, which is located away from the welded joint. At this location, there are no residual stresses to induce ratcheting. It is also observed that the fatigue cracks in all experiments occurred at the weld toe location where the ratcheting strain is the largest. This indicates that the fatigue life of materials is reduced in the presence of ratcheting. There are several observations made in the tests that need further exploration. Ratcheting in the tests occurs under displacement-controlled loading cycles, which seems kinematically inadmissible. The recorded load responses demonstrate cyclic softening for stainless steel, which is known as a cyclically hardening material. It is also observed that the socket-weld which is fabricated using the quarter-circumferential welding method has a longer fatigue life than the one which is fabricated using the full-circumferential welding method. Also, the socket-weld with four weld passes has longer fatigue life than the socket-weld with three weld passes. This study with its limited time and resource could not explore these issues. More elaborate experimental and analytical programs need to be developed to address these issues. In the analytical study, a finite element scheme for simulation of the welding residual stresses is developed. Thermo-elasto-plastic analyses, to replicate the welding processes of butt- and socket-weld piping joints and thereby to determine the residual stresses, are performed using ANSYS and ABAQUS. The numerical scheme developed in this research is first validated using the temperature and residual stress data in the literature. The simplified method for residual stress determination, through uncoupling of the thermal and residual stress analyses, is found to work reasonably well. When the analysis method is applied to the welded piping joints, high tensile residual stresses at or above the yield stress level, are calculated near the weld toe area, especially at the weld start/stop location. Analyses also indicate that the post weld heat treatment (PWHT) does not reduce residual stresses completely from the welded piping joints. The residual stress distribution does not change much when the slip-on gap in the socket-weld joint is reduced to zero. Fabrication of piping welded joint by quarter-circumferential welding yields a favorable residual stress distribution compare to the full-circumferential welding passes. Less distortion and an increase in fatigue life are observed in the earlier method of welding. Moreover, in the quarter-circumferential welding it does not matter much if the last welding pass is on the socket side or the pipe side. In case of the full-circumferential welding, the location of the last pass has been shown to influence the fatigue life. Fatigue response analyses demonstrate that the residual stresses do not relax completely after a few of inelastic cycles as assumed in the design methods. Because of the deficiency of the constitutive model, it was not possible to simulate the measured ratcheting responses in the welded piping joint experiments. In order to improve the fatigue design methods for welded joints, it is essential to develop a constitutive model that can simulate its ratcheting-fatigue failure responses. This, in turn, would require a carefully planned experimental program for developing a systematic set of welded-joint fatigue response data and temperature-dependent material response data.
Date: 2003-02-11
Degree: PhD
Discipline: Civil Engineering

Files in this item

Files Size Format View
etd.pdf 5.830Mb PDF View/Open

This item appears in the following Collection(s)

Show full item record