ABSTRACT
The aim of structural optimization is to achieve the best possible design. Using deterministic concepts this has been interpreted in the following manner (Moses, 1973). The objective function, load conditions, design requirements, failure modes and design variables are all treated in a nonstatistical fashion. Furthermore, it is assumed that design codes provide adequate safety factors to cover any likelihood of failure. By using automated optimization procedures, member sizes are found so that the objective function (e.g. cost, weight) is minimized without violating any design requirements. In some instances deterministic optimization promotes structures with less redundancy and smaller ultimate overload margins than obtained with more conventional and conservative design procedures (Feng and Moses, 1986b; Frangopol and Klisinski, 1989a). Consequently, deterministic optimized structures will usually have higher failure probabilities than unoptimized structures. The reason for this is that the safety factors specified in design codes usually refer to the design of elements, such as beams and colums, and do not allow for a simultaneous occurrence of many failure modes in an optimized structure designed to its limits (Moses, 1973). Therefore, a balance must be developed between the safety needs of the structure and the aims of reducing its cost. Clearly, this requires the use of reliability-based design concepts in structural optimization.
