Estimation of Drilling Rate Index Using Artificial Neural Networks and Regression Analysis

Abstract

In underground development gallery and tunnelling operations, the drillability properties of rocks have been studied by researchers in recently. Efficiency in excavation and drilling operations mainly depends on the success of mine production processes. Therefore, it is necessary to know the drillability properties of the formation to be excavated or drilled. This information can be obtained by detailed and costly field experiments. In this study, it is investigated whether the drillability properties of rocks can be determined rapidly and reliably depending on the brittleness index of the rocks. Brittleness index is a coefficient used in many mining designs. There exist a number of equations in literature to compute brittleness index of rocks. In this study, a new equation has been proposed for brittleness index as Bnew= \(\frac{\sqrt{{{}_{c}}^{2}+{{}_{t}}^{2}}}{\sqrt{2}{}_{t} }\). Effectiveness of this equation has been tested using linear and multiple regression models and has been compared with other brittleness equations in literature. In addition to Bnew index, effect of uniaxial compressive strength, tensile strength, three other brittleness equations, shore hardness and density variables are examined on drilling rate index value of rocks. Univariate regression, multiple regression and artificial neural networks are employed to estimate drilling rate index using these variables. Results have shown that using second degree multiple regression models and artificial neural networks drilling rate index can be estimated effectively.