Mechanical properties are interesting in many fields such as civil engineering, geotechnics, geomechanics and georesources exploration. Uniaxial compressive strength (UCS) is an important mechanical property determined to characterize natural and artificial rocks. This paper highlights the ability of artificial neural network (ANN), as an accurate and revolutionary method, to predict UCS within carbonate rocks. Thus, ANN was used to estimate the UCS values of the tested samples. For experimentation, we carried out ultrasonic measurements on cubic samples before uniaxial compressive strength, perpendicularly to the stress direction. The model was performed to link porosity, density and ultrasonic velocity to the UCS measurements. The resulted model would allow the prediction of carbonate rocks UCS values, usually determined with laborious experiments. Results confirmed that this model can be used as an economical and simple method to predict the uniaxial compressive strength of carbonate rocks.
Building materials are of important interest worldwide. Their lack in several regions and the high cost of their importation create the need to develop new technologies to facilitate their exploration and thus the estimation of physical and mechanical parameters that control carbonate rocks quality (Wang et al. 2012). In fact, determining physical and mechanical properties of heterogeneous materials is important in order to judge their future usefulness (Maghous et al. 2009). Mechanical, physical and geotechnical characteristics of rocks are evaluated through the measurement of parameters such as the Uniaxial Compressive Strength UCS, porosity, density, resistance to abrasion and fragmentation, etc.
Uniaxial compressive strength (UCS) is an important factor in determining material’s quality for mining, geological and geotechnical applications (Bieniawski 1974; Abdelhedi et al. 2020; Kurtulus et al. 2012). In rock mechanics, determining UCS is essential for tunnels and dams design, rock blasting, mechanical rock excavation, slope stability studies and other applications. Actually, the UCS assay via classical laboratory standards is laborious, time consuming and expensive. That's why the elaboration of prediction models as an indirect method for the estimation of such parameter is a field of research of a great importance (Yagiz et al. 2012, Ferentinou and Fakir 2017). Consequently, predictive models targeting specific parameters are emerging as an effective alternative method in all areas of scientific research. To determine the UCS, two methods are available.
The first is the direct method, in which the specimens are tested in the laboratory, and the second uses predictive models (indirect methods) (Baykasoğlu et al. 2008), recommended by many researchers for UCS predictions (Mohamad et al. 2015).