Using non-destructive tests for estimating uniaxial compressive strength and static Young’s modulus of carbonate rocks via some modeling techniques

Abstract

Prediction of elastic mechanical properties such as uniaxial compressive strength (UCS) and static Young’s modulus (Es) is one of the main purposes of studies in geological, geotechnical, geophysical, petroleum, and rock engineering projects. The UCS test is applied to determine them. However, this test is destructive, expensive, time-consuming, and requires high-quality samples. Therefore, using indirect methods seems to be indispensable for estimating dynamic elastic properties. Consequently, the main intention of this study was to predict the relationship between UCS and Es with dynamic poisson ratio (ϑd) and dynamic Young’s modulus (Ed), using simple and multivariate regression analysis (SRA and MRA), an artificial neural network (ANN), and support vector regression (SVR), and to compare and evaluate these methods with each other. For this purpose, different intact limestone rock samples of Asmari formation (ranged from limestone to marl) were collected from five different dam sites located in the southwest of Iran. Following regression analysis, the best equations for estimating UCS and Es of these samples with high and acceptable accuracy in terms of coefficient of determination (R2) and root mean square error (RMSE) are suggested. These equations are simple, practical, and accurate enough to apply for prediction purposes at preliminary design stages. Although ANN and SVR models are both powerful techniques, the SVR run time is considerably faster. Besides, when cocmparing the three models used, the SVR model was found more desirable and advantageous.