Coupled electrothermal-mechanical analysis for MEMS via model order reduction

This paper investigates model order reduction (MOR) techniques that can be used in conjunction with finite element schemes to generate computationally efficient solutions for multiphysics MEMS simulation. The Lanczos and Arnoldi algorithms are implemented to extract low dimensional Krylov subspaces from the finite element discretized system for model order reduction. A deflation procedure is employed in both algorithms to improve the solution convergence of the implicit iterations together with stopping criteria to automatically determine the reduced model size. Reduced order electrothermal-mechanical models are generated for a MEMS microgripper using the developed programs. A Guyan-based ANSYS substructuring analysis of the same device is also performed. Results discussion on all three techniques including preservation of full scale model properties such as dynamic behavior and stability are presented along with comparisons of reduced and full model simulation. The developed programs automatically generate compact structure preserving models and can be used to significantly improve the computational efficiency without much loss in accuracy and model stability for coupled-field MEMS simulation.