An advanced approach integrating methods to estimate hydraulic conductivity of different soil types supported by a machine learning model

Abstract Hydraulic conductivity is a critical parameter in geotechnical and hydrogeological engineering, influencing groundwater flow, structural stability, and environmental remediation. Traditional methods for estimating hydraulic conductivity are often time-consuming and costly, relying on empirical formulas that are limited to specific soil types. This study proposes a novel approach using an artificial neural network (ANN) to predict hydraulic conductivity for both fine and coarse-grained soils based on grain size distribution. A comprehensive dataset of 205 soil samples was used to train and validate the ANN model. The model’s performance was compared with traditional methods, including empirical formulas and multiple linear regression, to achieve realistic estimation. The ANN model demonstrated superior accuracy in predicting the hydraulic conductivity with a regression coefficient ( R 2 ) exceeding 92% and a root mean squared error of 0.077. A key innovation of this study is the development of a unified prediction approach that requires only the weight percentages of soil components, making it practical and cost-effective for engineering applications. The results highlight the potential of ANN to provide accurate and efficient estimates of hydraulic conductivity, offering significant advantages over conventional methods. This research contributes to advancing the use of machine learning in geotechnical engineering and provides a robust tool for soil characterization and groundwater management.