The prerequisites for fatigue analysis of tubular joints are knowledge of local stresses at critical areas and an appropriate design S/N curve. Cumulative fatigue damage can then be calculated for a given loading spectrum.
This paper describes an experimental investigation of an axially loaded small-scale T-joint to measure the maximum stress concentration and compare it with formulae from the literature, to obtain constant-amplitude fatigue data for comparison with proposed design curves and to investigate the suitability of Miner's rule from the results of variable-amplitude fatigue tests.
With the expansion of offshore oil exploration to hostile waters such as the North Sea there is an urgent need for reliable fatigue analysis procedures to be included in design codes for offshore installations. The tubular welded joint is the most widely used constructional detail in steel jacket type structures, and is an economical method of febrication, which produces a joint with a remarkable reserve static strength1. It has been used extensively without unduly large failure rates in environments, such as the Gulf of Mexico, in which the high risk of damage tends to be confined to short periods of storm conditions when the static strength of the joints is critical. In other locations where the wave spectrum includes appreciable numbers of waves that are higher than approximately 20 per cent of the design wave height, there will be a high risk of fatigue cracking. The complex stress distribution in a tubular joint, which is caused by the interaction of the restraint of the welded joint and the deformation of the tubular members, gives rise to high stress concentrations, adjacent to the weld toe, at critical positions around the joint. These areas of local high stress are potential sites for the initiation of fatigue cracks and progressive failure.
The fatigue analysis procedures proposed by the UK Department of Energy2 and the British Standards Institution have the same basic format as the AWS Fatigue Criterion which has been considered to be applicable to tubular joints1.
A designer using these procedures would have to carry out a suitable experimental or computational study, which takes account of geometrical parameters and modes of loading to establish stress concentration factors for each joint of the structure. He would then use the resultant local stress spectra at the hot spots in a cumulative damage calculation, based on an appropriate design SIN curve, to estimate the fatigue lives of the joints by Miner's rule.
An alternative technique of analysis, which can be used for chord members of some simple T- and K-joints, is based on the punching shear stress range. The nominal axial and bending stresses in the brace are used to calculate the spectrum of punching shear stress, Vp, using the formula 4
(Mathematical equation available in full paper)
