Assessing the unconfined compressive strength of lime stabilized soil blended with iron ore tailings using probabilistic and artificial neural networks approach

Abstract

Black cotton soil (BCS) mixed with Lime and Iron ore Tailings (IOT) was evaluated through laboratory experiments. BCS treated with 0, 2, 4, 6, and 8% Lime; 0, 2, 4, 6, 8, and 10% IOT by dry weight of soil, followed by testing considering different energies. A regression model equation was developed after data testing, on laboratory determined results for probabilistic analysis using MiniTab R15 version, which was used as limit state equation, and integrated into a F`ORTRAN program, to predict safety indices (SI) for unconfined compressive strength (UCS), in the array of coefficient of variation (COV) between 10–100%. Artificial Neural Network platform (ANN) was also carried out on same data set. The UCS increased with increase in both additives and elapse time.Reports from statistical study revealed a strong relationship between the UCS values and the soil independent parameters (lime L; iron ore tailings IOT; elapsed time ET; liquid limit LL; plasticity index PI; maximum dry density MDD; Optimum moisture content OMC, as well as Compactive effort CE) with determination coefficient R 2 = 87.7%. SI values for PI (0.314 to 0.519) and MDD (0.0574 to 0.285), demonstrated more effect on the UCS values, with wide range variability in their SI values, followed by ET (0.259 to 0.347) and L (0.259 to 0.455) while LL (0.318 to 0.336), IOT (0.31 to 0.324), and OMC (0.318 to 0.319) demonstrated the least effect, with slight variability in their SI values with change in COV. Results of ANN recorded determination coefficient R 2 = 97.23%, best validation performance of 12121.034 at epoch 10. Although, the documented findings did not meet the least regulatory value of 1.0 as mentioned by the Nordic Committee on Building Regulations at 10–100% COV. The obtained SI for all the independent parameters recorded higher values when placed in comparison to the untreated soil, which indicate significant improvement of strength properties of the soil. Results obtained for treated soil at 2 hours elapse time can use to model soil strength properties for dams or embankment applications under Heavy compaction energy.