Automation of the pre- and postprocessing of analyses performed by the nonintrusive coupling between Abaqus and an in-house code

Resumo

Structural design requires precise analyses that represent the actual behavior of the structure, ensuring safety and project optimization. In this context, many industry software, such as Abaqus, use the Finite Element Method (FEM) and perform well in simulations of large structures. However, they face difficulties achieving accurate results in simulations involving localized phenomena, such as cracks, holes, or discontinuities, which require high computational costs. To overcome the limitations of commercial software, a non-intrusive coupling strategy is used between Abaqus and the INSANE (INterative Structural ANalysis Environment) computational environment, based on the Generalized Finite Element Method with Global-Local Enrichment (GFEMgl). This strategy divides the analysis into three solution scales: global scale, which represents the entire structure, disregarding the effects of local phenomena, and is solved in Abaqus with a coarse mesh using FEM; local scale, modeled in INSANE, accurately simulating the region where phenomena occur with high discretization; and mesoscale, which facilitates the exchange of information between the scales. The mesoscale and local scale are solved using the global-local strategy applied to GFEMgl, while the coupling between the global scale and the mesoscale is performed through a procedure called Interactive Global-Local (IGL). For preprocessing, a Python script is used that automates the creation of the INSANE input file based on the structure modeled in Abaqus. Initially, the structural model is built in Abaqus, then the script is executed in the Abaqus interface within the mdb file, requesting the necessary information from the user for INSANE and the number of interfaces. Finally, using specific commands from the Abaqus documentation, the model data is automatically extracted and organized in XML format.Additionally, another Python script merges the Abaqus results into the INSANE postprocessing file, allowing a complete representation of the analyzed structure behavior. The xml file generated by INSANE, containing the final mesoscale result, is used to incorporate the results obtained in Abaqus. These automations represent an advance in the practical application of the strategy in the industry, making the process less prone to errors and more accessible and feasible for designers.