Fatigue crack growth studies on welded joints with multiple weld toe cracks

This PhD study examines fatigue crack growth analysis and testing case studies of multiple surface cracks for assessing fatigue cracks propagation life of welded joints with surface cracks at weld toe regions. Fatigue analysis was performed based on fracture mechanics concepts using the crack propag...

Full description

Saved in:
Bibliographic Details
Main Author: Tsang, Kin Shun
Other Authors: Pang Hock Lye, John
Format: Thesis-Doctor of Philosophy
Language:English
Published: Nanyang Technological University 2021
Subjects:
Online Access:https://hdl.handle.net/10356/153419
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:This PhD study examines fatigue crack growth analysis and testing case studies of multiple surface cracks for assessing fatigue cracks propagation life of welded joints with surface cracks at weld toe regions. Fatigue analysis was performed based on fracture mechanics concepts using the crack propagation driving force parameter of stress intensity factor (SIF). SIFs were obtained for plain plate and welded joint specimens with semi-elliptical cracks. XFEM was employed for improving the model mesh generation and insertion of crack front sub model, allowing for an approximation of cracks without the need for the mesh to follow the crack as in conventional crack modelling techniques, significantly simplifying the modelling of cracks. The importance of partitioning of the model and orthogonality of the crack in XFEM was demonstrated, and the SIFs obtained by XFEM were compared to benchmark solutions. The Mk factor solutions were used in a fatigue life prediction model to predict the growth of fatigue cracks at weld toe regions and compared with experimental fatigue test data. A multiple crack recharacterization method incorporating SIFs and Mk factor solutions was developed as a MATLAB program for fatigue crack propagation. S-N curve tests and FE modelling were conducted for welded pipe cut-out specimens, cruciform joints and rail welded specimens. For welded pipe cut-out specimens, an improved fatigue life analyses for multiple cracks was developed by considering the crack coalescence stage and fatigue crack closure. The prediction achieved a better fit to the tested fatigue life S-N curve results. An algorithm was developed for multiple fatigue crack propagation and demonstrated for a large number of initial starting weld toe surface cracks (12 to 22). The analysis provided details of the fatigue growth history of each crack, their sequence of coalescence and the evolving crack shape development (a/c vs a/t) to a dominant single, final crack before fracture. For cruciform joints, both tension loading and four-point bending tests were carried out to understand weld toe failures. XFEM was used to model a semi-elliptical surface crack to obtain SIFs and Mk factor solutions, which showed good agreement with established solutions. A procedure for fatigue crack growth and life prediction for cruciform joint was developed similar to that used for the welded pipe cut-out specimens, and the prediction was close to the tested fatigue life for tension loading. The algorithm of the multiple fatigue crack propagation for cruciform joints demonstrated the predicted fracture surface with 13 initial cracks presented the most detailed and best approximation of the experimental fracture surface, as compared to 5 or 1 crack. For rail specimens, inverted four-point bend tests were conducted to determine the fatigue strength of rail welded specimens. The results showed that the fatigue life of a specimen in the inverted fatigue test was about 3X that of the conventional test. From the fracture surfaces, it could be seen that the crack initiated from the head-to-web transition area along the edge. SIF solutions of rail specimens were developed based on 3D XFEM Model, and Mk factor solutions were obtained. A more powerful scheme to account for crack propagation based on SIF known as the coupled root mean square (RMS) method proposed by Cruse and Besuner was introduced and implemented. The fatigue life prediction provided a good fit to the experimental fatigue test data, and the predicted fracture surfaces were shown to closely resemble the actual fracture surfaces.