Adaptive Quadrilateral Refinement Considering Multiple Quantities of Interest

Abstract

This paper shows the performance of adaptive quadrilateral refinement of shells considering multiple quantities of interest. The adaptive refinement process is so developed as to have NASTRAN as the analysis tool, a posteriori error estimation in quantities of interest as the error estimator, and a quadrilateral refinement on a parasolid geometry as the mesher. The process intends to provide engineers an estimation of the quality of their linear static and modal analyses with respect to the quantities of interest, and an effective adaptive quadrilateral refinement tool for shells. For linear static load cases, the quantity type can be selected among three displacement components, six stress components and the Von Mises stress. The location of the quantity can be defined by specifying a node ID, an element ID or the global coordinates. For modal analysis cases, the quantities of interest are simply chosen from available natural frequencies. Up to five quantities of interest can be selected in one analysis and they may reference different loadcases from NASTRAN. Two numerical examples, a 2-D bracket and an automotive roof, are employed to demonstrate the performance of the process.