OasisLMS

Abstract: This study focuses on the prediction of properties of soils encountered during exploratory drillings using Measurement While Drilling (MWD) in the State of Illinois. MWD is an instrumented drilling technique where drilling parameters such as drilling rate, rotational speed, thrust, and torque, are recorded continuously. Advancements in MWD technologies have transformed the energy sector but its applications in geotechnical engineering remain limited due to a lack of systematic data-driven investigations to link MWD parameters with the geomaterial properties. This gap motivated this study to perform multiple exploratory drillings using an instrumented drill rig to measure the MWD parameters and use the measured data to predict soil properties. Exploratory drillings were performed at five different sites across the state of Illinois to record a robust database having variable soil properties. Data analysis revealed missing torque values for a fraction of data, and this was addressed using machine learning based imputation technique. After imputation of missing data, machine learning analysis was performed to train the XGBoost algorithms for prediction of soil type, SPT-N, and Qu values of the encountered stratigraphy. Validation of the trained XGBoost models exhibited satisfactory coefficients of correlation of 0.98, 0.93, and 0.96 for the prediction of soil type, SPT-N, and Qu, respectively. The outcomes of this research will help in the development of a comprehensive dataset for future MWD studies and reinforce MWD as an economical supplementary method to conventional site investigation techniques.

TUGCE BASER

International Conference Proceedings