The effect of plate thickness on fatigue strength of transverse non-load carrying fillet welds in axial tension was investigated. Plate thicknesses were in the range 12.5–80 mm. The main part of the investigation was concerned with joints in which the weld throat and the transverse attachment plate were scaled in proportion to the thickness of the load-carrying plate, i.e. the welds were geometrically similar, typical of load-carrying welds. The effect of plate thickness was found to follow a power law, with a decrease in fatigue life by 80% when increasing the plate thickness from 12.5 mm to 80 mm. This corresponds to a decrease in fatigue strength of 40 per cent. The experimental results were analyzed by fracture mechanics. The agreement between experiment and analysis was good. A supplementary test series with constant thickness of the load-carrying plate and varying weld size and attachment plate thicknesses showed that these parameters have only a minor effect on fatigue strength as long as the dimensions are reasonably in scale with the load-carrying plate thickness. A survey of relevant data confirmed the findings of the present work, and design recommendations were evaluated on this basis.
Current fatigue design codes and recommendations for welded steel structures 1,2,3,4 are based on S-N data obtained from laboratory tests with plate thicknesses typically in the range 10-30 mm.5 Many structures, however, are generally designed with much larger thicknesses. This has given rise to a concern about the effect of plate thickness on the fatigue capacity of welded joints, and a possible unconservatism in the present codes.6 In fact, code revisions have been announced, with a thickness factor which reduces the allowable stresses when the plate thickness is increased.7 Although these revisions have not yet appeared officially, certifying agencies have started to take the effect of plate thickness into account in their approval procedures for offshore structures.8
It should also be mentioned that there is a general recommendation on methods for the assessment of the significance of defects in welded joints, in which plate thickness is considered. In addition, one code contains an optional fatigue design procedure based on a simplified fracture mechanics analysis, in which there is an allowance for the thickness effect.4 None of the two latter procedures are applied in regular design.
The present study was undertaken in order to provide data for the effect of plate thickness, and to establish a fracture mechanics model for the observed effect.
In Ref. 10 a preliminary account of the present investigation was given, covering plate thicknesses up to 60 mm.
In the present work, results are reported for the plate thickness range 12.5-80 mm, and for various weld geometries. The experimental data are analyzed by a fracture mechanics model, and the results are compared to proposed design procedures.
