Relating gradational parameters with hydraulic conductivity of sandy soils: a renewed attempt

Criticism has been made in the recent literature on the classical predictive equations of hydraulic conductivity (k) of sandy soils for being solely dependent on the smallest or effective grain size. In this study, a renewed attempt has been made to correlate k with different gradation sizes and gradation parameters of sandy soils. One hundred eighty sandy soil samples having natural and composite formations are tested in the laboratory for gradation and hydraulic conductivity, out of which, 130 and 50 samples are used for model development and model validation, respectively. Data analyses show that a reasonable gradation and hydraulic conductivity spectrum are considered in the model development phase of the current study. The linear D₁₀-k relationship shows the highest goodness of fit than the linear relationship among k and other grain sizes from D₉₀ to D₁₀. Response surface and multiple linear regression-based correlation analyses are conducted which reveal that incorporation of different gradation parameters, sizes, and compaction parameters along with D₁₀ can enhance the health of correlation for the prediction of k-value. Hence, 6 response surface analysis-based and 1 multiple linear regression analysis-based models are proposed in the current study to predict k-value using different gradation parameters, gradation sizes, and compaction parameter as the predictors. The variations between the experimental and predicted values by proposed models are found to be in a range of ±12 and ±10% for the database used for model development and validation, respectively. An independent data set is also used to validate some classical predictive equations and proposed models in this study; the former equations are found to overestimate the k-value in comparison to the latter ones.