Classification of the intact carbonate and silicate rocks based on their degree of thermal cracking using discriminant analysis

摘要

Discriminant analysis is a multivariate statistical tool that generates a discriminant function to predict about the group membership of sampled experimental data. In this study, discriminant analysis was performed using IBM SPSS software package (version 23) to discriminate between pre-defined groups of measured dynamic properties of thermally treated selected carbonate and silicate rocks. The range of temperature variations was selected from 35 degrees C (laboratory temperature) to 200 degrees C to estimate the change in dynamic properties including Q-factor (Q), resonance frequency (F-r), elastic Young's modulus (E-d), damping ratio (xi), and specific damping capacity (psi) by using the "Erudite Resonance Frequency Meter" apparatus in accordance with ASTM C215 test procedure. The test results revealed significant variations in the values ofQ,F-r,E-d,xi, and psi with the rise in temperature. The measured dynamic properties of selected sedimentary rocks were classified into 2 discrete groups based on their extent of thermal treatment (i.e., 35-100 degrees C and 100-200 degrees C). Discriminant analysis-based classification results showed the sensitivity level of 86.70% and specificity level of 100.00% between predicted and original group membership. The estimated model hit ratio of 92.00% found better than maximum chance criterion and proportional chance criterion that indicates the high level of significance of classification results. Q-statistic results also satisfied the high prediction accuracy of discriminant function. The outcomes of this study could provide useful references in the classification and characterization of experimental data related to geotechnical and geomechanical studies.