Caracterização do efeito de enrijecimento por tensões e identificação de cargas em estruturas baseada em respostas dinâmicas

Abstract

It is known that stress State can influence, to a large extent, the static and dynamic behavior of structural systems such as columns, frames, plates and shells. Such phenomenon is known as stress-stiffening effect. This work presents a study on the stress-stiffening effect in relation with the dynamic characteristics of structures. The emphasis is placed on two different aspects, namely: the development of a modeling procedure enabling to characterized the influence of the externai loads on the dynamic behavior of two-dimensional frames and rectangular plates; the development of an inverse procedure intended for the determination of externai loads, given the dynamic responses of the loaded structure and a numerical model. With this aim, the Finite Element Method and the Assumed Modes Method are used to model the dynamic behavior of externally loaded two-dimensional frames and rectangular plates. Both types of modeling procedures are implemented in MATLAB® environment. The dynamic responses are characterized in terms of modal parameters (natural frequencies and vibration mode shapes) and frequency responses functions. The variations of the natural frequencies as a function of the externai loads are interpreted in the context of a buckling dynamic criterion. The influence of the stress-stiffening effect on the dynamic behavior of rectangular plates is also characterized through laboratory experiments performed on Steel and aluminum plates, to which different load scenarios are applied. The load Identification procedure consists in solving a constrained optimization problem in which the cost function expresses the differences between the measured and model-predicted natural frequencies and/or the vibration mode shapes of the loaded structure. The externai loads, assumed to be unknown, play the role of design variables. The optimization problem is solved numerically using both a classical gradient-based and a pseudo-random algorithm known as Particle Swarm Optimization. The results obtained confirm the paramount influence that the stress-stiffening can have on the dynamic behavior of structural systems and demonstrate the effectiveness of the modeling procedures and the feasibility of the load Identification procedure based on the dynamic responses.