Fatigue evaluation of welded details – using the finite element method

Doctoral thesis, 2013

The fatigue evaluation of welded details is generally based on the notion of nominal stress, using the classified S-N curves with corresponding fatigue classes for typical details. An approach of this kind should be used with extra caution to ensure that the load effects for components are accurately captured, because an ever-increasing number of welded details are resulting in a limited number of possible treatable design cases. The fatigue of welded structures is a somewhat complex and progressive form of local damage which can be evaluated more accurately using local failure approaches, such as the hot-spot and effective notch stress methods. Methods based on fracture mechanics can also be applied in cases where fatigue cracks are detected or can be assumed to exist. A large number of welded details with complex geometry and load conditions that are known to be critical with respect to fatigue can be found in welded steel structures. Estimating more detailed and accurate information on the stress state of these details is very difficult without using finite element analysis. On the other hand, the result obtained from finite element analysis can be highly sensitive to the modelling technique, as the stresses obtained from the local failure approaches are often in an area of high strain gradients, i.e. stress singularities. In the first part of this thesis, the fatigue evaluation of welded details using the finite element method was studied to evaluate the applicability and reliability of the local failure approaches for details that are typical in steel and composite bridges. In order to obtain a better understanding of these methods in terms of implementation and limitation, both “simple” and complex welded details were studied using various finite element models. In the second part of this thesis, the fatigue evaluation of welded details in existing steel structures was investigated in order to examine the fracture resistance of fatigue-cracked welded details and to obtain more reliable inspection periods for the cracked structures. The effectiveness, accuracy and applicability of the crack-propagation analysis based on the linear-elastic fracture mechanics for distortion-induced fatigue cracking in bridge structures, were investigated by performing several crack-propagation analyses. The results obtained in this thesis show that the local failure approaches provide better fatigue life estimations in comparison with the conventional methods, even though these approaches require more effort for modelling. In addition, the fatigue assessment of critical details susceptible to distortional cracking can be performed more accurately using the local failure approaches. This type of detail should not be estimated using the conventional nominal stress based methods due to the highly localised deformation whose effect cannot be captured by these methods. On the other hand, the crack-propagation analyses showed that the crack growth rate decreases as the crack length is extended, due to relaxation, as the web gap becomes more flexible because of the extended crack. The fracture mechanics-based methods can be utilised in order accurately to determine the required inspection period and retrofitting technique.