Machine Learning to predict a ship's fuel consumption in seaways
Paper i proceeding, 2016
Fatigue cracks can be observed quite frequently on today’s ocean crossing vessels. To ensure the safety of ship structures sailing in the sea, it is important to know the residual fatigue life of these damaged ship structures. In this case, the fracture mechanics theory is often employed to estimate how fast these cracks can propagate along ship structures. However, large uncertainties are always associated with the crack prediction and residual fatigue life analysis. In this study, two uncertainties sources will be investigated, i.e. the reliability of encountered wave environments connected with shipload determinations and different fracture estimation methods for crack propagation analysis. Firstly, different available codes based on fracture mechanic theory are used to compute the stress intensity factor related parameters for crack propagation analysis. The analysis is carried out for both 2D and 3D cases of some typical ship structural details. The comparison is presented to illustrate the uncertainties of crack propagation analysis related with different codes. Furthermore, it is assumed that the structural details will undertake dynamic loading from a containership operated in the North Atlantic. A statistical wave model is used to generate wave environments along recorded ship routes for different years. The uncertainties of crack growth analysis related with encountered weather environments is also investigated in the study. The comparison of these two uncertainties indicated the requirement of further development for the fracture mechanics theory and associated numerical codes, as well as the reliable life-cycle encountered weather environments.