Multi-objective design of post-tensioned concrete road bridges using artificial neural networks

Abstract

In order to minimize the total expected cost, bridges have to be designed for safety and durability. This paper considers the cost, the safety, and the corrosion initiation time to design post-tensioned concrete box-girder road bridges. The deck is modeled by finite elements based on problem variables such as the cross-section geometry, the concrete grade, and the reinforcing and post-tensioning steel. An integrated multi-objective harmony search with artificial neural networks (ANNs) is proposed to reduce the high computing time required for the finite-element analysis and the increment in conflicting objectives. ANNs are trained through the results of previous bridge performance evaluations. Then, ANNs are used to evaluate the constraints and provide a direction towards the Pareto front. Finally, exact methods actualize and improve the Pareto set. The results show that the harmony search parameters should be progressively changed in a diversification-intensification strategy. This methodology provides trade-off solutions that are the cheapest ones for the safety and durability levels considered. Therefore, it is possible to choose an alternative that can be easily adjusted to each need.