Quasi-Optimization for Efficient Stiffness Design of Structures

Abstract

Different promising developments in structural optimization have been made, with some obstacles for more widespread and general application laying mainly in some lack of generality and flexibility of computer software and the relatively high computational effort envolved. This is especially true when problems are governed by considerations in structural dynamics, which is the case for structural design of most spacecraft structures where often stress constraints are relevant only for local load introduction areas. In addition, minor changes in the requirements (i.e. in the constraint levels) often occur especially in the earlier stages of the development process and can make a “strict” optimum obsolete. But at the same time it are these earlier development stages where at least the overall structural design is frozen. So computational very efficient approaches are desirable, which must not necessarily solve for the strict optimum and also not necessarily consider each detail in stressing which is usually left for the final design stages. The benefits of such “quasi-optimization” approaches then are simplifications in the solution algorithms where then not more than one or two reanalyses are required in contrast to a considerably higher number in “strict optimization”. This is achieved for example by approximations in both the physics and optimization leading then to linear, quadratic or simplified nonlinear optimization problems. Very stringent requirements often can no longer be reasonably satisfied by a passive optimal design approach, but then active control has to be applied.