Shape Optimization of Axisymmetric Solids Using Automatic Differentiation and Nurbs Parametrization

Abstract

Shape optimization of axisymmetric solids is performed in this work using a NURBS parametrization of the geometry coupled with a Sequential Quadratic Programming (SQP) algorithm where the sensitivity analysis is performed by automatic differentiation (AD). The structural analysis is evaluated by the Finite Element Method. The framework proposed is computationaly efficient and accurate and in addition remeshing techniques are also avoided. Large thickness variation are allowed with the methodology. As a result, structures with improved structural performance are obtained for linear elastic structural problems.