Topology Optimization for a Magnetic Actuator Using Different Gradient-Based Solvers

Abstract

This paper presents a comparative study on the performance of density-based topology optimization using different gradient-based solvers. Three common solvers are employed: the method of moving asymptotes (MMA), interior point optimizer (IPOPT), and sparse nonlinear optimizer (SNOPT). The gradient solvers are compared regarding the convergence behaviour, computational time, and solution quality, i.e. maximizing the attractive force, when topology optimization is used to design a magnetic actuator as a case study. The results provide valuable insights into the strengths and limitations of MMA, IPOPT, and SNOPT in solving topology optimization and offering guidance for selecting the appropriate solvers. The results proved that SNOPT is suitable for largescale problems as it has fast convergence and low computation time compared to MMA and IPOPT solvers. Further, IPOPT is not suitable for large-scale problems due to high computation time and large amounts of intermediate materials, i.e. regions where the density function differs from 0 and 1. Moreover, the quality of the optimal solution is only slightly affected by the used solver.