Statistical Analyses and Geotechnical Evaluation of Nubia Sandstone, Golden Triangle Area, Egypt

摘要

Great efforts had been made to use indirect non-destructive tests in the geotechnical evaluation of rocks, especially sandstones, employing different empirical equations. However, most of these equations have been derived from hard and compacted sandstones data;therefore, the focus of this research is on weak and weakly compacted sandstones, aiming firstly to obtain empirical equations for estimating their characteristics, secondly to demonstrate and visualize the correlations between the studied variables, and finally to cluster the studied samples based on their characteristics. To attain these aims, twenty oriented block samples were collected from Nubia sandstone, central Eastern Desert, Golden Triangle area, Egypt. These samples were prepared and tested according to standard test methods, including uniaxial compressive strength (UCS), Brazilian tensile strength (BTS), Schmidt rebound number (SRN), porosity (n), bulk density (ρ), and ultrasonic P-wave velocity (UPV). The loss on ignition (LOI) was also employed as a physicochemical test for classifying the studied samples and indicating pores filling materials. The results revealed that these sandstones are characterized mainly by high n, low ρ, and low UPV values and these give an indication of weakly compacted and weakly cemented sandstone with shallow burial diagenetic conditions. Based on UCS and elastic modulus values, these sandstones are mainly classified as very low strength and highly yielding rocks. The results of regression analysis show satisfactory correlations between physical and mechanical characteristics, indicating the suitability of obtained empirical equations to deduce these properties. Principal component analysis revealed that the LOI, BTS, SRN, and USC have a positive correlation to each other and weakly correlated with ρ and UPV, which positively correlated to each other and negatively correlated to n. The results of agglomerative hierarchical clustering revealed that the studied samples can group into three main clusters depending on their USC, LOI, and n values.