Abstract
Drilling operations for oil and gas wells are considered one of the highest operating costs for the petroleum industry, and hence, implementing the new technology-based systems is highly required for cost reduction and efficient functionality performance for the drilling system. This paper introduces a machine learning study for the role of composite lithology schemes on drilling rate prediction as it is one of the common practical challenges of developing machine learning models for predicting drillability rate. A field case study introduces artificial neural networks (ANN) and extreme gradient boosted trees regressor (XGBoost) machine learning models to develop and validate a drilling rate prediction. The dataset incorporates surface rig sensor parameters such as weight on bit, drilling rotation speed, pumping rate and pressure, torque (as input parameters), and drilling rate of penetration (as the predicted output parameter). Confirmed ground truth data, including lithological characteristics and formation tops, complements the dataset. Two wells’ data contribute to model development, while blind unseen well data validates the models. The training dataset encompasses complex lithology formations sandstone, dolomite, anhydrite, limestone, and interbedded shale. The methodology follows a comprehensive workflow covering data preparation, filtering and cleaning, statistical analysis, feature engineering, model development, parameter optimization, and accuracy assessment through coefficient of correlation, average absolute percentage error, and root mean squared errors. The results showcase the high accuracy of the developed machine learning models (coefficient of correlation exceeding 0.99) during training and validation while the blind testing showed R of 0.83 and 0.89 for XGBoost and ANN respectively. The study explores the impact of complex lithology schemes on drillability rate prediction, employing data analytics and machine learning models. Real-time drillability rate prediction, facilitated by this model, serves as a technical guide for optimizing drilling parameters, enhancing performance, and achieving optimal mechanical-specific energy.
