Abstract:

In this thesis, optimum design algorithms for composite steel I-girder and tied-arch bridge structures were developed. The design optimization algorithms determine the optimum cross sectional dimensions of bridge members under AASHTO-LRFD design provisions such that with these cross sections the bridge has the minimum weight. The determination of optimum cross-sectional dimension is carried out by selecting the steel profile designation from the list of available steel profiles in the case of a composite steel I-girder bridge and selecting the dimensions of the rectangular steel plates which are used to construct steel built-up sections for the members of a tied-arch bridge. It has been decided that, in tied-arch bridges ties, arch ribs, and bottom and top bracings are made of built-up box sections, whereas built-up I sections are utilized for floor beams and stringers. Bars are adopted for hangers. The mathematical modeling of the design optimization problems in both cases turns out to be discrete nonlinear programming problems. Metaheuristic algorithms are used to obtain the optimum solutions in both cases due to their efficiency in finding solutions to such programming problems compared to other optimization techniques. Five recent metaheuristic algorithms were selected in the development of design optimization algorithms. These include the artificial bee colony algorithm (ABC), the biogeography-based optimization algorithm (BBO), the exponential big bang-big crunch algorithm (EBB-BC) and the symbiotic organisms search algorithm (SOS). The fifth algorithm is the enhanced artificial bee colony algorithm (EABC), which is an enhanced version of the standard ABC and is proposed in this thesis. Design optimization algorithms require the responses of bridges under several design load arrangements and the construction of influence lines. This is achieved by using the application-programming interface (API) facility of SAP2000. The bridges were modelled in SAP2000 as three-dimensional modeling and the design algorithms were coded in MATLAB. The algorithms benefit from the application programming interface process available in SAP2000-API to obtain the analysis results. It was found that the developed optimum design algorithms were able to optimize the bridges and reduce their weights by 13.06% and 13.20% for the composite steel I-girder and tied-arch bridges respectively. It was also found that the proposed EABC is efficient and better than the standard ABC in terms of the convergence history and providing the best results.