Drilling Rate Index Estimation via Soft Computing Techniques

Rock drillability is an important factor in mining, construction, and geotechnical engineering, among other industries. The drilling rate index (DRI) is a commonly used drillability evaluation method. Typically, DRI values are determined via laboratory testing using brittleness (S20) and Sievers’ J miniature (SJ) drilling tests. However, specimen preparation for these drilling tests is a laborious and time-consuming process that can make it challenging to determine DRI values. This study attempted to estimate DRI values based on mechanical rock properties. The mechanical rock properties include the uniaxial compressive strength (UCS), Brazilian tensile strength (BTS), and brittleness indices (B1–B5) as a function of UCS and BTS. Datasets were taken from relevant literature. A predictive model was developed using symbolic regression and machine-learning techniques. Symbolic regression uses mathematical functions and operations to determine the most appropriate mathematical equation that best describes the relationship between input variables and the DRI in a dataset. Several machine-learning techniques were used, including random forest, decision tree, k-nearest neighbors, and support vector machine. The results showed that the DRI values of rocks can be estimated in a fast, practical, and feasible manner using a symbolic regression model.