Structural Optimization of Bridge Cantilever Decks : Applications of an Automated Design

Abstract: Civil engineering projects involve great investments and great impacts. For that reason,engineers have a commitment with an efficient and optimal use of resources. Researchers inuniversities claim that a lot could be achieved by applying structural optimization into realprojects, even though this approach has not gained the same popularity in the industry over thelast decades.The purpose of the present thesis is to explore the possibilities offered by structural optimizationand to verify its applicability in realistic and complex structural engineering problems. Amongthe questions regarding design optimization, it was emphasized feasibility, efficiency and userfriendliness. The chosen structural system was a bridge cantilever deck. The analysis was limitedto the transversal design of the structure and the goal of the optimization was to reduce investmentcosts. In order to guarantee efficiency of the simulations, a "longitudinal length convergence"analysis was performed. It consisted of determining the minimum required longitudinal length(perpendicular to the cantilever length) that ensured reasonable accuracy. The purpose of thisanalysis was to reduce the computational time during the optimization process. In order toautomate the analysis, MATLAB was used in connection to Abaqus (to perform the FE Analysis).There were three different sets of results presented: the length convergence, application toreal projects and parametric study. In the first application, it was shown that the requiredlongitudinal length (lx) proportionally decreased as the cantilever length (lc) increased. It wasalso observed the presence of the edge beam implied in consistently larger longitudinal lengthsfor the same accuracy tolerance. With respect to the second application, two projects wereconsidered and the structural optimization presented alternatives with significant investment costreduction in a reasonable time. Furthermore, it was observed that a design solution without theedge beam reduced the costs even more. Finally, the parametric study confirmed that the costreduction obtained by eliminating the edge beam was not restricted to only certain cantileverlengths. Furthermore, it was possible to obtain the pattern of thickness variation as function ofthe cantilever length.The results of this research suggest that structural optimization could be an alternative totraditional design methods used today in consulting offices and its possibilities transcend puredesign achievements.