CAD-based adjoint optimization using other components in a CAD model assembly as constraints

This paper presents a CAD-based shape optimization process which exploits the capabilities of modern CAD systems to enforce assembly constraints within the optimization process. The assembly constraints are imposed using adjacent components in the CAD model assembly, which the component being optimized is not allowed to overlap with. This is important in industrial workflows, where unwanted interference can often result during the final product assembly. Here, an optimization framework is presented where the parameters defining the features in a feature-based CAD model are used as design variables, and their gradients are computed by combining design velocities with sensitivities computed using adjoint methods. The benefits of this framework are three-fold: (1) the use of adjoint methods makes the computational cost essentially independent of the number of design variables, (2) the optimized geometry is available as a feature-based CAD model that can be easily used for downstream processes, (3) the optimized geometry respects space constraints imposed by other parts in the assembly. In this paper, the developed framework is demonstrated for the optimization of models created in CATIA V5, to be assembled with other components defined in the CATIA V5 assembly workbench.