Discrete Element Analysis And Identification Methods Of Ice Loads On Ship Structure

Ice-going vessels play a significant role in polar scientific expeditions,navigation,and resource exploitation.However,as major environmental loads during navigation in the polar region,ice loads can influence the structural safety of which can cause severe structural damage or fatigue damage to the ship structure.Additionally,ice loads on the ship structure constitute the foundation of a navigation warning mechanism and an optimal design of ship structure.But the ship-ice interactions are complex given the numerous shape parameters of ship hull,various forms of ship-ice relative movements,and the anisotropic characteristics of sea ice mechanics.Subsequently,ice loads differ considerably in aspects including time-course characteristics,periods,distribution characteristics,and statistical laws.The natural distribution of sea ice exhibits a strong discrete characteristic and the sea ice is in a continuous failure state under the action of ship structure.Then the discrete element method(DEM)can be used to establish a numerical model for ice load analysis and the bond model in DEM can be used to simulate the breaking process of sea ice.Recently,DEM has undergone rapid development in terms of the contact model,searching method,and bond model.Thus,DEM is widely used in the analysis of ice loads on marine structures in cold regions.The full-scale measurement of an ice load on a ship is another important approach to analyse the ice loads on ship.The research focuses on the reasonable measurement technique,feasible load identification model,and statistical analysis of ice loads.The study focused on the numerical model of ice load analysis based on DEM and full-scale ice load identification.The main aims of the study are as follows:(1)A numerical model for sea ice based on DEM was established.Subsequently,the breaking process of sea ice was simulated via the parallel bond model and generalised contact model(GCM).The GCM with local contact points was adopted to simulate the gradual fracture process of sea ice.Uniaxial compression tests and three point bending tests on sea ice were simulated via GCM.The study examined the effects on the macro-mechanical parameters of sea ice via key micro parameters in GCM,including the distribution of the local contact points,elastic modulus,ratio between shear stiffness and normal stiffness,and strength parameters(normal bond strength,shear bond strength,and contact frictional term).(2)The DEM was used to analyse the ice loads on ship structures.The interaction between a ship and level ice was simulated.Subsequently,global ice resistance in the simulation was validated via semi-empirical formulas.Broken ice fields with different concentrations were generated.Subsequently,the effects of concentration on the extremes and periods of ice loads were analysed.(3)The core techniques of the full-scale ice load monitoring system were analyzed.The ice thicknesses and ice-induced accelerations along the route of a polar area were counted.The optimal layout of strain sensors was determined via the analysis of the induced signals measured in the full-scale and in the impact test of a shell structure.A load identification test on the plate was performed to validate the Influence Coefficient Matrix method(ICM),which is typically adopted in ice load identification.(4)An ice load identification model was constructed based on the Green kernel and regularization methods.The ice loads on the ship hull are mainly identified via ICM,which is based on statics.However,ice loads are typical impact loads and the identification method should consider the dynamic effect.In the study,the Dirac function was adopted to establish the Green matrix.Subsequently,the singular value decomposition theorem and Picard criteria were used to analyse the ill-posed characteristic of the load identification system.Different regularization methods and the corresponding regularization parameter optimization methods were adopted to improve the recognition effect in the ice load identification.Experiments and numerical simulations of load identification on the shell structure were performed to validate the proposed model.(5)An far-field ice load identification model of ship structure was constructed based on support vector machine.The strain sensors are difficult to install,however,at the required locations on the hull because the watertight compartment is cramped.In this situation,the general ice load identification methods are not applicable.Additionally,the far-field load identification method can be applied to overcome the aforementioned restrictions with regard to the installation location.To determine the nonlinear relationship between ice loads and ice-induced strains in far-field load identification,the least square support vector machine algorithm(LS_SVM)and Gaussian radial basis function kernel were adopted.The experimental applications were performed to validate the feasibility of the LS_SVM procedure and a full-scale ice load identification model was established.The identified ice loads on the ship hull in the present study can reflect the characteristics of ice loads in time domain.Analysis of a fitting function,analysis of a variation coefficient,and parametric analysis of the ice loads measured in full-scale were performed.Finally,the main contribution of the thesis was concluded and future topics were suggested.