Research On Numerical Methods For Structural Failure Simulation

Structural failure has attracted much attention in recent years, especially public buildings with large span. When their bearing capacity is exceeded or local failure appears causing by an emergency, a wide range of progressive failure or collapse will take place, which threatens people’s lives and property at any time. Previous works on structural failure analysis mainly use one-dimensional elements such as beams or trusses. Basically they meet the demands of macroscopic analysis, however, microscopic deformation features of the structure are ignored and complicated components, for example joints and beams with special-shaped section, cannot be analyzed included in the structure. Therefore, the true structural failure mechanism and bearing capacity cannot be achieved. In order to avoid the non-accuracy of one-dimensional elements, from a microscopic perspective, this dissertation research on structural failure based on finite particle method (FPM) using refined elements such as shell elements or solid elements.After a review of previous researches on structure failure analysis both at home and abroad, the methods, advantages and problems facing this dissertation are discussed and the research framework is also presented.Some basic concepts of FPM and fracture mechanics are introduced, including point value description, path element, deformation description, governing equation, cohesive element, cohesive zone model, fracture energy, spoil stress, Traction-separation law (TSL) etc. As for the extrinsic cohesive zone model, the insertion of cohesive elements will result in a change of the topology connection between bulk elements, hence a novel method is proposed to handle this topological modification and it has great generality for arbitrary dimension. In order to show how they cooperate with each other for structural failure analysis, the computation procedures are also given, which lays an important foundation for the following chapters.Without a loss of generality, the fracture of ordinary 2D solid (plane stress) is conducted firstly. Based on FPM, the formulations of a triangle 2D element are derived, including governing equation for particle movement and element internal forces, which expands the element library of FPM. Also, a quadrilateral cohesive element is derived to cooperate with it for fracturing 2D solid. Some numerical examples are verified to demonstrate the feasible of this method.Then, from 2D to 3D solid, the formulations of a tetrahedron 3D element are derived. Additionally, material nonlinearity is also considered rooted in Mises elastic-plasticity model. A trigonal prism cohesive element is derived to cooperate with it for fracturing 3D solid. Several numerical examples are given to show the efficiency of this method.As an expansion of the element library of FPM, a kind of thin-shell element is developed, which is superposed by two parts, CST membrane element and DKT plate element. Compared with other isoparametric high order shell element, it has simple computational process and high efficiency of developing. Different from 2D and 3D solid, the crack through the thickness of thin-walled structure is implicit, therefore, a cohesive element with an equivalent opening consisting of both tension and bending based on releasing the correct fracture energy is adopted to model the fracture of thin-walled structure.Besides fracture, buckling is also a typical structure failure mode. In engineering practice, the critical bearing capacity of structure must be calculated accurately for security. As for buckling analysis, FPM often applies the external force on the structure step by step, that is, the external force is always increasing during the analysis. Obviously, there is a limitation for this loading strategy, when the structure is buckling and the bearing capacity goes down, the increasing external load will not adaptively decrease to capture the full equilibrium path. Consequently, the explicit arc length method is adopted to combine with FPM. Adaptively loading or unloading is the essence of this combination. Aimed at contact problem during buckling or large deformation, a contact algorithm is modified for explicit analysis and extended from’particle to surface contact’to’beam to beam contact’.Finally, as an application of the work in this dissertation, a typical tube-truss structure is modeled with refined thin-shell elements and the complete failure process, from buckling to microscopic fracture, is actually achieved, which indicates that the approach in this dissertation for structure failure analysis is feasible and reliable. In order to simulate the fracture of weld joint in engineering, a multi-scale FPM model is also establishd, which consists of shell and solid elements. Since the refined numerical model has a large scale of computation, Kinetic damping and parallel computing techniques are employed to accelerate the calculation and a new parallel scheme for assembling particle internal forces is proposed.From theoretical consideration to algorithm implementation, numerical examples suggest that the computational theory presented in this dissertation for structural failure analysis is feasible and has great stability and generality. It can be taken as a new effective analysis tool for engineers and researchers. The conclusions and problems that should be studied further are summarized at the end of this dissertation.