New rock strength-based UCS-V[formula omitted] correlation equation for different rock types by statistical regression methods

Abstract

Uniaxial compressive strength (UCS) is an essential parameter for the characterization of rock strength. P-wave velocity (Vp) measured using ultrasonic testing has been commonly used to predict UCS because this method has many advantages over conventional uniaxial compressive tests, such as its simple operation, low cost, and non-destructive nature. Vp can also be used to calculate dynamic elastic moduli. A Vp-based equation that can be used to evaluate the UCS of numerous types of rocks with high accuracy, however, remains a challenge. Effects such as porosity, weathering, and mineralization have been seldom considered although they significantly affect the UCS-Vp correlation, especially in weak rocks. This research proposes a rock strength-based UCS-Vp regression equation for engineering purposes, which can be used to predict the UCS of different rock types that have a Vp over a critical value and that have been classified by strength (high or low). The training data on which this equation is based yield a coefficient of determination of 0.96 and a standard deviation of 18.1 MPa, and the mean absolute error of the regression model is 14.7 MPa after testing. A critical Vp value of 3.5 km/s is used to evaluate how a rock is affected by porosity; if the rock Vp is greater than this critical value, a more in-depth strength analysis is needed. The critical rock strength judgment involves several hammer blows being applied to a rock sample to fracture or chip it. This determines whether the rock sample has a UCS above or below 100 MPa and low or high porosity, and thus the variable input values for the prediction equation. The UCS-Vp correlation is linear for rocks with greater than critical strength and logarithmic for rocks with lower than critical strength. The new model provides a quick and effective way to evaluate UCS for engineering projects and significantly extends the usage of the UCS-Vp correlation equation to different rock types while maintaining a high degree of accuracy.