Return to Research Projects

Reliability-Based Optimal Design of Steel Box Structures

Principal Investigator(s) 

A steel box structural design has been traditionally used for bridges and surface marine vessels and is on the basis of engineering experience/judgment and deterministic approaches, such as working stress design. Recent developments in reliability-based design have lead to the development and application of the load and resistance factor design (LRFD) approach, in which inherent uncertainties are accounted for in an indirect manner. Structural optimization algorithms and tools have also seen an increased application in the area of structural design in the context of weight reduction, cost savings, and other design improvements. A three-step strategy for the design of such structures was developed, which combines the innovative aspects of both reliability-based methods and structural optimization theories and algorithms. After selecting an initial design as a starting point, the methodology then improves on this through deterministic optimization, in which design variables, objectives, and constraints are defined in a purely deterministic manner. Concerns regarding major components, selected configurations and layouts, loading, and design objectives/constraints can be addressed at this point and the process can be repeated if necessary. This deterministically predicted optimal design is then used as the basis for reliability-based optimization, which is the final step of the methodology, in which design variables, objectives, and constraints are defined in a probabilistic context. Through its ability to account for uncertainties, a reliability-based framework greatly complements any deterministic prediction. The methodology has been applied to the design of two typical surface vessels as examples: one simple and one complex.