Mineralogical Identification of Clays Using K-Mean Clustering

The degree of expansivity of clay depends on the mineral(s) present; therefore, identifying mineral(s) in clay is essential to assess its swelling and shrinkage characteristics. Fine-grained soils could be classified as Kaolinitic or Montmorillonitic where Montmorillonitic soils are relatively more expansive in nature. The expansive soils are very problematic as they affect the stability of structures found on them. X-ray diffraction (XRD), differential thermal analysis scanning electron microscopy (SEM) etc. techniques could be able to predict the mineral(s) in clay with high accuracy; however, employing such techniques in soil investigation is not possible due to their sophistication and handling of bulk heterogeneous soil mass. Many researchers and codes suggested the expansivity of soils based on index properties such as liquid limit, plastic limit, shrinkage limit, etc. This study aims to identify the Kaolinitic, Montmorillonitic and Mixture of both soils by applying an unsupervised learning clustering technique namely K-Mean clustering.