Multiobjective Optimization of Bridge and Viaduct Design: Comparative Study of Metaheuristics and Parameter Calibration

Abstract

This article focuses on multiobjective optimization in the design of bridges and viaducts. The problem is characterized as a multi-objective optimization, with the objective functions being the construction cost, the environmental impact (CO2 emissions) and the design service life. For optimization, three commonly used metaheuristics in structural optimization problems were tested: Multiple Objective Particle Swarm Optimization (MOPSO), Nondominated Sorting Genetic Algorithm II (NSGA-II) and Strength Pareto Evolutionary Algorithm 2 (SPEA2). The results showed that MOPSO outperformed the other methods, achieving the highest hypervolume and Pure Diversity values. Once the best performing metaheuristic was defined, the calibration of the MOPSO parameters was then developed to improve the quality of the solutions found, reduce the execution time and increase the robustness of the algorithm. Using the Taguchi method with an orthogonal matrix consisting of 54 experiments, five parameters were evaluated at three different levels, totaling 270 analyses. The results indicated that parameter calibration led to an increase in the average hypervolume compared to the results before calibration. Moreover, the coverage of two sets revealed the superior performance of the calibrated set, demonstrating better trade-offs among the objective functions.