Nonlinear Modeling Of Mechanical Joint Interfaces For Structural Dynamics Application

The existence of complex multi-scale, multi-physics and nonlinear behaviors on joint interfaces are mainly response for complex dynamics of assembled structures. The dynamics of joint interfaces is one of the most important branches of structural dynamics. Modeling of mechanical joint interfaces is also a challenging scientific problem, due to the complexity of interface behaviors and difficulties of direct experimental observation.There are mainly two categories of methods proposed to model the mechanical joint interfaces. The first class of methods are called “from down to up”, which start with the micro-scale physics of asperity contact. Then, the statistical method or fractal model is applied to model the multi-scale behaviors of rough interfaces. The second class of methods are called “from up to down”, which start with the analysis of the macro-scale dynamic response of assembled structure. Then, nonlinear system identification is used to model the joint interfaces. The work in this thesis are summarized as follows:1) The elliptic curve interpolation is applied to approximate the average contact stress of asperities in the mixed elastic-plastic deformation phase. The contact model of rough surfaces is formulated using probability and statistics theory.2) The relations of the tangential displacement with restoring force and interface energy dissipation is derived from the classical MINDLIN solution. Then, the KD roughness function is used to yield the contact formulations of rough surfaces.3) With the assumption of coulomb friction for asperity contact, the probability distribution density of critical slip force is derived based on the KD contact equations. Then, the modeling principle of IWAN model is used to deduce the contact model of rough surfaces.4) The empirical mode decomposition is used to get the nonlinear characteristics behaviors of joint interface from the macro-scale dynamic responses of jointed structures. The BP neural networks is applied to get the model parameters of joint interfaces.5) The method combining the empirical mode decomposition and fast-slow analysis is applied to identify the nonlinear characteristics of jointed structures, such as time-variant stiffness and damping. Then, the reduced-order model of jointed structure is developed and verified by experimental result of bolted joint beam. The works of this paper will build the fundament of further researching of nonlinear structural dynamics of joint interfaces.