Soil Profile Stratification Based on Cone Penetration Test Results Using k-means and Hierarchical Clustering

Abstract

The design of the foundation for any structure depends mainly on the soil stratigraphic profiling and the characteristics of the soil. Accuracy of the profiling and identification of the boundary of layers help create accurate and cost-effective designs. In recent years, cone penetration test (CPT) is used widely in soil stratigraphic profiling since it produces near-continuous records for both the cone and sleeve resistance. It is used widely for soil profiling and engineering soil characteristics, especially in important projects such as those related to petroleum production projects. This paper applies the k-means method of clustering and hierarchical clustering to the soil behavior index obtained from the cone penetration test to identify the soil profile for the soil. These two approaches have been applied to the results obtained from the sound of CPT tested on the soil at the site of Nasiriyah province in Iraq. Applying these two methods is very useful as it can detect the layers and identify the soil layers simply. The analysis shows that the soil at the site can be classified into two main layers with several secondary layers. Collecting the clustering of the soil behavior index with the other engineering properties can give a strong imaginary picture of the soil stratification. The study uses different methods for verifying the results obtained from the hierarchical and k-means clustering method such as inconsistency and silhouette graphs. A comparative study has been achieved on the obtained results by the suggested approaches with the adjacent borehole logs showing very good agreement.