Pré-Processador para modelos reticulados e plano métodos elementos finitos

Abstract

This Master's Thesis describes a pre-processing program, which is part of the INSANE (INteractive Structural ANalysis Environment) system, for the creation of Finite Element models. With it is possible to combine bar, plane state and plate elements. The software, implemented in JAVA, was developed according to the object oriented programming paradigm and has a layered architecture based on a combination of three design patterns: the pattern Model-View-Controller (MVC) that allows the independence between the model and view layers; the patternObserver whith, through a mechanism of changing propagation, allows the synchronization of the model with various pairs\view-controller"; and the pattern Command which encapsulates the actions of the program in separate classes in an organized way. This segmentation favors the expansion of the program and the creation of new commands. To allow di_erent discretizations of the same geometric model, the preprocessor was divided in two modules: GEOMETRY and MESH. The first one allow the creation and edition of points, lines, curves and regions. The second provides resources for the discretization of geometric objects and their attributes definition. On the regions, meshes of plane state and plates elements can be generated through the Transfinite Mapping (Lofting, Bilinear or Trilinear). The geometric model and the mesh model data are stored in data structures for planar subdivision which uses the half-edge concept. It allows the storage in an organized way and establishes the relations of adjacencies, making it possible a fast access to the data. Euler operators are used to create and to manipulate the topological entities that compose the model (vertices, edges and faces) and to represent it in the data structure. The use of multiple views, combined with the application of geometric transformations and projections, allows diferent visualizations of a model and more than one model can be simultaneously observed. The models can be serialized as JAVA objects or XML files.