Evaluating the influence of petrographic and textural characteristics on geotechnical properties of some carbonate rock samples by empirical equations

Abstract

The effect of petrographic and textural characteristics on geotechnical properties of some carbonate rock samples including limestone and travertine was investigated using Texture Coefficient (TC) and regression analyses. For this purpose, nine rock block samples were collected from quarries and road trenches in northern and northwestern parts of Damghan, northern Iran. Physical, index and mechanical properties namely specific gravity, dry and saturated unit weights, porosity, water absorption, slake-durability index, Schmidt rebound hardness, P-wave velocity, uniaxial compressive strength, point load strength index, Brazilian tensile strength and block punch strength were determined in the laboratory. Petrographic, mineralogical and textural investigations were studied by thin section and X-ray diffraction methods. Texture coefficient and required parameters were determined by JMicroVision (v1.27) software. Regression coefficient (R) was obtained between “0.10 and 0.98” by simple regression analysis. Three variable map shows that TC of the studied rocks is controlled by the presence of major minerals including quartz and calcite. Good direct linear relationships were found between TC and percent of calcite and quartz with high correlation coefficient (R = 0.79 and 0.85). The regression analyses indicated good correlations between TC and engineering properties, especially between γdry, γsat, HS, UCS, IS(50) and BPS. Also, no good relations were found between TC and n, Wa and VP. Statistical coefficients including R, RMSE, VAF, MAPE and PI were calculated to assess performance and validity degrees of obtained equations and the regression analyses. Performance appraisal shows the model of BPS and TC has a higher performance than the other models. Experimental and calculated values of geotechnical properties that obtained from laboratory tests and predicted by statistical models were compared with 45° line (y = x). Based on the results, the trend lines of E, HS, UCS, IS(50) and BPS models are more fit to y = x line and shows high validity of experimental models. Results revealed that the texture coefficient is a useful parameter for predicting geotechnical properties of the rocks.