Automatic shape design of double-arch dams using k-means algorithm

Abstract

Dams are super-structures widely used in water conservancy engineering fields for several uses. Their long-term safety is a focus of social concern, and it is strictly correlated to the design layout. In this paper, parameters’ inter-correlations for the design layout of double-arch dams were analyzed. From 37 Spanish real dams, 296 parameters have been filtered and collected. These values, mainly regarding geometrical dimensions, have been divided into 8 categories and combined with each other. A total of 192 numerical analyses have been carried out by using a k-means algorithm that can be considered an artificial intelligence (AI) technique to support the human limitations in managing and analyzing several parameters, for instance, the heights, lengths, thicknesses, and volume of dams. Preliminary results provided a new relation between the concrete volume and height of the dam. Results provide disaggregated values where each parameter is correlated with another one. It appears cluster 1 provides a better calibration. This allows us to understand their weight and effects on design layout. This research provides not only a new approach but also practical values for more accurate analyses.