Characteristics of Designs and Load Paths in Structures Optimized for Robust Damage Tolerance

Abstract

Structural optimization for robust damage tolerance couples non-linear progressive analyses with sampling based reliability analyses which makes it computationally expensive. Understanding design features and characteristics of alternate load paths in robust damage tolerant structures can result in efficient methods for optimization. In this paper we present the optimization of an 18 bar truss structure with random distributed damage and investigations of its design features and alternate load path characteristics. The paper presents computational challenges in fitting accurate Kriging surrogate models for the robust damage tolerance objectives. The optimum robust damage tolerant design obtained is compared to other optimum designs. It is shown that redundancy by itself does not guarantee damage tolerance. In order for it to be damage tolerant the structure must be tailored to take advantage of alternate load paths provided by the redundancy. Alternate load paths that arise and progressive failure responses are different for energy absorbing structures and damage tolerant structures that must sustain load after damage. However, for both designs the progressive failure response indicates bifurcations that lead to a small number of discrete final ultimate load values, even though the initial damages were random. The linear correlation of damage to the ultimate load response was used to further eliminate some branches in the progressive failure collapse that lead to reduced load capacity after damage. Finally, a method to quantify the number of alternate load paths is presented.