Spectral Fatigue Analysis of Plate Surface Hot-Spots: A Practical Solution to the Stress Direction Issue

Spectral Fatigue Analysis using coupled hydrodynamics and finite element models has now become a common practice for the fatigue strength assessment of offshore units, with established procedures given in Classification Rules. However, users are facing a practical issue that is almost never mentioned in the procedures. Indeed, many fatigue hot-spots are located on a plate surface, as opposed to plate edges. For such hot-spots, the finite element model results are the three components of the plane-stress stress tensor. Therefore, the outcome of the Spectral Fatigue Analysis is a set of three transfer functions (RAOs). On the other hand, our industry’s practice regarding the fatigue strength model is still the proven « design S-N curve » approach in combination with the Palmgren-Miner’s damage summation. As a consequence, today the engineer is left with no clear instruction about the proper way how to close this gap between the three stress RAOs on the one hand, and the single stress S-N curve on the other hand. If any advice is given, it is most often to consider the principal stresses, tentatively extending to spectral analysis the classification rule load cases approach. However, principal stress determination is a non-linear procedure that is not compatible with spectral analysis in frequency domain. Turning the spectral results into time domain to overcome this limitation is extremely costly and is not straightforward. Of course, a rational solution to this issue would be the adoption of a multiaxial fatigue damage criteria in lieu of the uniaxial S-N curve. But until such a multiaxial fatigue criteria is widely accepted in our industry, users have to square the circle, and force their stress tensor RAOs into the existing rule criteria. In this paper, a practical solution to reconcile plane stress results and conventional S-N curve criterion in spectral fatigue is proposed: the “facet approach “.