Ship Longitudinal Strength Modelling

The aims of the research work presented in this thesis are to assess the effects of weld-induced residual stresses and initial imperfections on the collapse behaviour of a ship's hull girder, and to demonstrate the close correlation that exists between the hull's ultimate longitudinal strength and the maximum load-carrying capacity of its components under compressive loads. A theoretical method for evaluating the ultimate longitudinal strength of a ship's hull girder under vertical bending is described. The hull's midship cross-section is discretised into structural elements such as stiffened panels, plate elements and hard corners. Effects of buckling of compressive components are allowed for by incorporating the load-end shortening curves of unstiffened and stiffened plates into the hull strength analysis. An incremental approach is employed to derive the moment-curvature relationship and hence the peak moment for the hull girder. Curvature, rather than bending moment, increments are imposed on the hull girder to enable the post-collapse behaviour to be followed. Comparisons with tests on welded steel box girder models and other analytical methods show that the agreement is satisfactory. Prior to determining the vertical moment-curvature relationship for a hull, the load-end shortening curves need to be established for each stiffened panel forming the hull's mid-ship cross-section. An analytical method is developed to examine the large deflection elasto-plastic behaviour of stiffened panels under uniaxial compression. The method is based on a beam-column approach in which the longitudinally stiffened plating is treated as a series of beam-columns formed by the stiffeners and an associated width of plates. The beam-column model is continuous over three supports provided by transverse frames to take effect of interaction between adjacent spans into account. Dynamic Relaxation is employed to numerically solve the non-linear equilibrium equations. The strength contribution from the plating, allowing for buckling effects, is accounted for by using the plate average stress-strain curves. The load is applied through end displacements such that both the pre- and postcollapse behaviour can be traced. It is shown by comparisons with test results and other analytical methods that the present one satisfactorily predicts the behaviour of stiffened compression panels. A numerical method to generate the plate average stress-strain curves for the stiffened panel analysis is proposed. The results from an existing parametric study on the large deflection elasto-plastic behaviour of practical plates in compression with constrained edges are used as basic data. A simplified procedure is followed to interpolate the average stress-strain curve for the plate with parameters different from the standard cases. An analytical study on the ultimate strength behaviour of longitudinally framed frigate-type hulls is presented. Five naval frigates designed in the 1950's and 1960's are analysed by the present incremental approach. The derived load-end shortening curves for the stiffened panels and moment-curvature curves for the hull girders are presented. It is shown that the ultimate strength of longitudinally framed hulls is strongly influenced by the full-range behaviour of components under compressive loads in association with bending. In particular, the ultimate hull strengths are closely correlated with the maximum load-carrying capacities of the critical stiffened panels which are located in deck structures in the sagging condition and in bottom structures in the hogging condition. Two simple expressions for predicting the ultimate moment capacity of longitudinally stiffened hulls in the sagging and hogging conditions are proposed. They are based on the results of the numerical analysis for appropriate initial imperfections in the plate and stiffened panels. The ultimate bending moments of a variety of hull and box girders predicted by the strength formulae are compared with the numerical and experimental results. Satisfactory agreement is obtained which suggests that the proposed formulations could form the basis of an improved design method.