Automatic Optimization of the Cross-Sections of Earth Dams and Embankments Using the ODACO AI-Based Optimization Program
Automatic Optimization of the Cross-Sections of Earth Dams and Embankments Using the ODACO AI-Based Optimization Program
Amin Rezaian1 Mohammad Davoodi2 Mohammad Kazem Jafari3
1) Assistant Professor, Department of Civil Engineering, Islamic Azad University, Gorgan Branch, Gorgan, Iran.
2) Associate Prof, International Institute of Earthquake Engineering and Seismology, Tehran
3) Prof, International Institute of Earthquake Engineering and Seismology, Tehran
IOI Code
:
ABFG-ACAGGK
Publication :
12th.international Congress on Civil Engineering, Architecture, Art and Urban Development(12icsau.ir)
Abstract :
most economical cross-sectional geometry of earth dams are among the computationally complex problems in geotechnical engineering. In this study, a computational framework implemented in an optimization program named ODACO is introduced for optimizing the cross-sections of zoned earth dams. Within this framework, the Ant Colony Optimization (ACO) algorithm is employed to search for the critical slip surface and determine the optimal geometry of the dam shell, while satisfying slope stability constraints under various loading conditions as well as construction constraints. Furthermore, an artificial neural network (ANN) has been trained using data obtained from several hundred two-dimensional finite element analyses to rapidly estimate the seismic response of dam cross-sections. The performance of the proposed framework is evaluated through two case studies: the Gotvand Olya Dam and the Alborz Dam. The results for the Gotvand Olya Dam indicate an approximately 13% reduction in the embankment cross-sectional area compared with the existing design. For the Alborz Dam, a reduction of 4.2% is reported. These findings demonstrate that combining metaheuristic optimization with an ANN-based surrogate model can make the preliminary design process of earth dams more efficient, optimized, and straightforward.