57288 AUTOMATED AND ERROR CONTROLLED MODEL REDUCTION FOR DURABILITY BASED STRUCTURAL OPTIMIZATION OF MECHANICAL SYSTEMS(Optimization and Sensitivity Analysis in MBS)

Abstract

Structural optimization based on durability values is an important tool for the design of light weight structures. One essential part of the optimization loop is the Elastic Multibody System (EMBS) simulation. An important issue for the simulation of the EMBS is the automated reduction of the flexible body's degrees of freedom. Therefore, totally automated Krylov-subspace based reduction processes for mechanical systems are introduced. Besides, an H_2-optimal SOIRKA algorithm is compared with a Second Order Adaptive Global Arnoldi (SOAGA) algorithm. For the automated reduction process error estimation is necessary. Three different error estimators are introduced and compared. In addition, automated error controlled reduction with snapshot-based approximation of frequency-weighted Gramian matrices is a suitable reduction scheme. Advanced basis construction schemes from Reduced Basis methods, e.g. the Greedy and the adaptive Greedy algorithm, improve and accelerate the reduction process. A numerical example demonstrates the approximation capability in the frequency domain.