Analysis of the effect of bending and torsion for fatigue in container ships

Abstract

Ocean-crossing vessels should be designed with sufficient fatigue strength. The highcycle fatigue principle with specific S-N curves is used in the maritime industry to predict the fatigue life of ship structures. For conventional ship structures, the stress range distributions are provided by classification societies and are mainly based on empirical experience. However, a ship may change its designated trade region, leading to a change in encountered wave environment. This will create a discrepancy between design and actual stress range distribution. Furthermore, for a novel ship design, data of fatigue loads is not available to guide the structural design. Consequently, so-called direct calculation methods are introduced in order to compute the loads and corresponding structural stresses. For the fatigue assessment of ship structures, the stresses are mainly caused by wave loads on the hull. These hydrodynamic loads can be computed using various theories and numerical implementations. As the method complexity increases, the computation may become more sensitive, leading to larger uncertainties. Moreover, usually, many different sea states with several operational conditions are considered, consuming much computational time in the early fatigue design stage. The objective of this thesis is to study the efficiency and fatigue result of using different computational methods as well as the effect of fatigue damage contribution from bending and torsion. Different methods, from strip theory to advanced non-linear panel methods, are employed in order to estimate the hydrodynamic loads on a 4,400TEU container ship. Subsequently, the structural stresses are computed using both finite element methods and engineering beam theory combined with different options for local stress concentration. The corresponding fatigue damage is then estimated using various spectral methods and compared to direct rainflow counting to investigate the scatter. Based on the results presented in this thesis it is concluded that a linear panel method for wave load analysis and engineering beam theory for structural stress calculation, combined with a simple spectral fatigue model, can give us accurate enough results and the most convenient/fast computation for simple details. However, at locations other than the mid-section, correct accounting for warping is required and an FE method is recommended.