Improving uniaxial compressive strength estimation of carbonate sedimentary rocks by combining minimally invasive and non-destructive techniques

Uniaxial compressive strength (UCS) is the most used parameter to measure rock strength. However, restrictions in sampling large volume of material, the need of very large set of results and onsite characterisation of UCS non-destructively are requirements in many scientific and engineering investigations. The estimation of UCS from a single non-destructive or minimally invasive technique (NDT) may result incomplete because each NDT is sensitive to different compositional and textural factors. This paper combines open porosity, P-wave velocity, Leeb hardness and micro-drilling resistance force to estimate USC for a wide range of carbonate sedimentary rock types with different petrographic characteristics. Results reveal that mineralogical composition significatively affects micro-drilling resistance force profiles and Pwave velocity values, especially for quartz-bearing rocks. In addition, texture controls substantially the reproducibility of tests sensible to rock surface properties, such as Leeb hardness and micro-drilling resistance force. Fifteen simple and multiple expressions for UCS are fitted. Linear expressions have shown better coefficients of determination (R-2) than non-linear equations because of the linearity shown by individual parameters. Curve fitting improves as the number of petrophysical parameters increase in the multiple linear regression analysis. The best correlation is found when the equation incorporates all the mechanical parameters obtained non-destructively as well as open porosity (R-2 = 0.910). Leeb hardness is always the most significant variable of the fitted regressions and its addition into multiple linear equations causes an increase of R-2. Open porosity also improves R-2 whereas drilling force and P-wave velocity have a lower statistical weight in the expressions. The UCS estimation from all NDT, without considering open porosity, shows a good correlation (R-2 = 0.899), which presents the advantage that they can be obtained non-destructively with portable equipment and can provide a numerous set of results at relatively low cost.