Using empirical correlations and artificial neural network to estimate compressibility of low plasticity clays

Abstract

Consolidation settlement tests are costly and time-consuming, whereas tests for determining soil physical properties can be performed very rapidly and at lower costs. Therefore, it will be very helpful if the soil compression index (CC) can be determined using soil physical parameters. This study investigated correlations between the CC and the physical properties of undisturbed and remolded Tehran clay by performing 125 consolidation tests and through determining the physical properties. Based on the results, the correlations between the CC and dry density (γd), between the CC and initial void ratio (eo), and between the CC and wet density (γw) are valid and have correlation coefficient (R2) of 0.87, 0.87, and 0.89, respectively, for undisturbed Tehran clay. These correlations have been proposed for engineering applications in this area. Furthermore, available empirical correlations were compared with those presented in this study and the results suggested that the low accuracy of some of the available correlations for estimating the CC of Tehran clay soil required accurate evaluations before using them in engineering applications. Moreover, using artificial neural network showed high potential in predicting the CC and have coefficient (R) close to 1.