Refining Compression Index Estimation for Fine Soils: Insights from Large Data and Sensitivity Analysis

The compression index of fine-grained soils is a critical parameter for determining structure settlement. In the laboratory, determining this parameter is time-consuming, costly, and laborious. Alternatively, determining this significant parameter based on soil index properties can be advantageous for preliminary design. This work compiled more than 1500 datasets to review and assess the valid correlation and, secondly, incorporate the index properties to better estimate the compression index of the fine-grained soils. This work indicated that previously established correlations, from which an individual index property provides the compression index values, cannot be applicable for many datasets and different types of fine-grained soil. Multivariable regressions were utilized to represent the actual properties of the soil and estimate the compression index. Concerning the analysis, models that incorporate the liquid limit, plasticity index, and initial void ratio can be utilized as good models. Different advanced mathematical models were employed to improve the accuracy of the model, including multivariable non-linear regression (MNLR), full quadratic (FQ), and full quadratic with high-order interaction (FQHOI). The result indicated that FQHOI can perform better than the other models. Finally, sensitivity analyses were conducted to highlight a parameter with the most influence on the compression index, and the results indicated that the initial void ratio had a superb influence on the value of the compression index.