Density gradient‐based adaptive refinement of analysis mesh for efficient multiresolution topology optimization

In topology optimization, the finite-element analysis of problem is generally most computationally demanding task solution process. order to improve efficiency this phase, in article we propose represent regions with zero density gradient by a coarser mesh. The design instead represented uniform We motivate gradient-based adaptive refinement discussing topological meaning and how it can help avoid loss information during projections or interpolations between meshes. also study adaptiveness mesh its ability detect change design. An posteriori error performed as well. Furthermore, provide theoretical numerical considerations on reduction number degrees freedoms respect case. This translates into faster analysis, show numerically. Finally, solve several test problems, including large 3D problems that parallel computer cluster, demonstrating applicability our procedure scale computing iterative solvers.