Experimental Investigation on Prediction of Physico-Mechanical Properties of Sedimentary Rocks Using Mechanical Parameters Obtained During Rotary Drilling

Abstract

The strength properties of rocks are frequently required during the introductory phase of rock engineering projects, including rock excavation, tunneling, designing of supports at underground mines, blast hole designs, etc. For the determination of rock strength properties like uniaxial compressive strength, tensile strength etc., at the laboratory, several rock core samples with high-grade quality is the prerequisites. Coring is an expensive process, and it is not always possible, particularly for weak and highly fractured, thinly bedded, foliated, and block-in-matrix rock mass. In addition, converting the core samples into test specimens as per the established standard is a tedious and time-consuming process. With all these constraints and problems, many times, the determination of strength or other rock properties may not be feasible in the direct method. Therefore, the subject of the indirect method used for approximation of rock properties has a wide scope. The main aim of this research study is to estimate the physico-mechanical properties of rock using the selected mechanical parameters obtained during the rotary-type rock drilling. In several rock engineering projects, either in construction or mining environments, rotary drilling operation is often widely used. If an estimation of some vital rock properties is possible during the rock drilling process, it may be a great advantage for engineers and geologists. In this experimental investigation, during the drilling of several rock blocks with different physico-mechanical properties, the mechanical drilling responses, such as thrust and torque at the bit-rock interface, are collected considering various drill operating parameters using a drill tool dynamometer. Similarly, at the same time, an acoustic parameter such as vibration data is also collected at the drill head using a sound/vibration data acquisition system (DAQ) with an accelerometer sensor. The response of these mechanical parameters collected during the drilling of various rocks was then analyzed and correlated with physico-mechanical rock properties. ii Prediction models (Type-I) for physico-mechanical rock properties such as UCS, BTS, SRN and density were developed using thrust, torque, including vibration data, using single order multiple regression methods. Similarly, prediction models (Type- II) were also developed using thrust and torque and excluding the vibration data, using second-order multiple regression methods. The prediction performance and validation of Type-I and Type-II are checked. The results showed that the Type-II model would predict the rock properties, i.e., UCS, BTS, SRN and density with less NRMSE than the Type-II model by 2.58%, 0.56%, 4.3%, and 3.17%, respectively. Similarly, compared to the Type-I models, the Type-II model would decrease the MAPE by 1.33%, 1.4%, 2.72%, and 0.24%, respectively. However, both types, i.e., Type-I and Type-II, could estimate the rock properties within 15% or acceptable errors. Due to the high sensitivity of vibration data to the spindle speed rather than by the UCS or other properties of rocks, and its high cost, it could conclude that the Type-II model might be useful, especially for estimating UCS and BTS at laboratory capacity without core samples. Besides, ANN models are developed for the prediction of rock properties. It was observed from the ANN models' prediction performance that the ANN models could estimate the physico-mechanical rock properties comparatively better than the Type-II model. In this case, NRSME of UCS BTS, SRN, and density were reduced by 1.97%, 7.38%, 1.74%, and 3.49 %, respectively. The results concluded the superiority of soft- computing models over the statistical models. In this study, strength the relationship between considered rock properties and drilling specific energy is also investigated. Initially, the average drilling specific energy is calculated for each rock type using Teale's equation. The average specific energy was 24.20 MJ/m3, 28.78 MJ/m3, 35.68 MJ/m3, 36.09 MJ/m3, 38.40 MJ/m3, and 42.12 MJ/m3 for shale (UCS=19.6MPa), sandstone-1 (UCS =37.5MPa), sandstone-2 (UCS=65.1MPa), sandstone-3 (UCS=72.4 MPa), limestone-1 (UCS=95.3 MPa), and limestone-2 (UCS=119.2MPa), respectively. The drilling specific energy of each rock type was then correlated with their physico-mechanical rock properties. The results iii showed that the strength of the relationship of UCS, BTS, SRN and density with drilling specific energy is good with R = 0.948, 0.892, 0.859, and 0.908, respectively.