The prediction of compressive strength of travertines with special reference to laminae-orientation using index tests

In this research, different experimental techniques have been used to determine the strength of four types of travertines, selected from Hamedan and Markazi Provinces, west and central parts of Iran respectively. For this purpose, after sample preparation and assessment of mineral properties, the selected samples were subjected to physical and mechanical tests with special reference to two major anisotropic angles (i.e. perpendicular (┴) and parallel (∥)) between bedding/lamination planes and the major loading directions. Based on the test results, it can be concluded that the percentage/type of matrix and porosity have a more important effect on the physical and mechanical properties than the rock structure. Statistical analyses, including simple and multiple linear regressions, were applied to identify those physico-mechanical parameters that are more appropriate to predict uniaxial compressive strength (σc┴ and σc∥ ) and to establish some new equations. Data analysis in simple regression shows that bulk specific gravity (saturated surface dry, SSD) and Brazilian tensile strength are the most and the least influential factors on σc┴ and σc∥ , respectively. Based on multiple regression method, equations with only two parameters (a physical and a mechanical parameter) were extracted for calculating σc┴ and σc∥. Among all proposed equations, models 1 and 6 based on SSD-Vp and SSD-CPI variables, seem to be reliable and agreeable to predict σc┴ and σc┴∥, respectively. The performance of multiple regression models are evaluated by comparing statistic parameters, including correlation coefficient (R), root mean square error (RMSE), mean error (ME) whose values for σc┴ are 0.86, 11.81, 0.11 and for σc∥ are 0.86, 12.97, 0.36. Also based on t-test and F-test, all the proposed models in multiple regression, show significant correlations with σc at α-level = 0.1.