Resumo:
The majority of steel structures in engineering, as well as a number of mechanical components, are constructed using different welding processes. On the other hand, the phenomenon of mechanical fatigue is the cause of approximately 80% of mechanical failure, being the sole damaging process or acting in combination with other processes. Nevertheless, the analysis of fatigue behaviour of welded joints is usually a complicated matter due to the different influencing factors resulting from the nature of the joint, and is generally realised in cuasi-empirical form because of the non-existence of models and/or theories that would allow to predict their behaviour.
Several methods to estimate the fatigue life of welded unions exist at present, including among others the nominal stress approach, the hot-spot stress approach, the notch stress approach, the local strain approach and simplified methods based on fracture mechanics.
The majority of these methods is standardized in literature or accepted in international recommendations. Nevertheless, they are based mainly on the use of Wöhler curves or S-N curves which require as a design parameter a nominal or equivalent stress applied to the analyzed configuration. By using these nominal parameters is not possible to study the influence of different geometric, mechanical and microstructural parameters that form part of the definition of the fatigue behaviour of a welded joint because these diagrams only are indexed according to the joint configuration and only differentiate between steel and aluminium. In addition they use empirical corrections to account for some variables like plate thickness, reinforcement angle, etc.
Due to its technological importance a major drawback of the present methods is that they do not allow to predict the fatigue behaviour of the joint and only provide conservative nominal values of the fatigue resistance for a given configuration not taking into account many of the influencing variables.
Recent works have provided the bases for new procedures of fracture mechanics analyses that allow to analyze the problem thoroughly by considering models that allow to study the mechanism of associate damage including the different variables that govern it. The present thesis circumstantiates the development of these models and the verification of their ability to explain and to predict fatigue behaviour of welded joints. The fundamental motivation of this work is based on the feasibility of relying on methods, theories, and models that allow to predict the fatigue behaviour of a certain configuration and the quantification of the influence of the different affecting variables without realizing experimental testing of the joints. The technological importance associated with the application of these theories and models is certainly significant.
At first, the thesis presents a summary of well-known models together with an analysis of their disadvantages and limitations. Then, an analysis of the new tendencies and the theories, models and approachs arisen in the last decade is given, and an integral fracture mechanics based model is presented, defining guidelines of applications and simplifications for the quantification of the necessary properties for its application. It is followed by a detailed analysis of specific configurations of butt welded joints, highlighting the importance of the stage of early crack propagation for the definition of the fatigue behaviour of welded joints without defects.
Afterwards, an analysis of results recently published by international researchers is presented and finally some important concepts associated to the new proposed fracture mechanics based approach and to the applied and/or implemented models, theories, techniques and procedures are summarized emphasizing on the importance of some conclusions and on evidence that shows the potential of this approach.
