Exploring the Link between Surface and Downhole Drilling Data for Real-time Evaluation of Mechanical Specific Energy in Slightly Deviated well

Mechanical Specific Energy (MSE) is a key concept in drilling engineering that is used to quantify the energy required to drill a unit volume of rock or formation. It serves as a crucial parameter for evaluating drilling efficiency and optimizing the rate of penetration (ROP). The primary drilling parameters influencing MSE include downhole weight on bit (WOB), downhole revolutions per minute (RPM), torque on bit, and flow rate. While MSE in vertical wells can be estimated using surface-measured values of these parameters (except torque), this approach does not apply to horizontal and deviated wells due to discrepancies in surface readings. These inconsistencies present significant challenges in real-time MSE estimation. Although downhole drilling parameters related to MSE can be directly measured using Measurement While Drilling (MWD) tools, real-time applications are hindered by sampling delays and data processing constraints. To address this, several models have been developed over the past decade to estimate downhole parameters based on surface drilling data. However, these models often rely on predefined coefficients, which can limit their accuracy and adaptability across different well conditions. This thesis aims to evaluate the accuracy of existing models which are used for estimating downhole drilling parameters and propose modifications by integrating additional factors to enhance prediction accuracy. Furthermore, the study explores the application of machine learning techniques in predicting downhole drilling parameters, with the goal of improving MSE evaluation. The findings of this research indicate that existing models struggle to estimate downhole parameters accurately, particularly in deviated wells. However, the modified models proposed in this study demonstrate improved accuracy. Additionally, machine learning techniques provide promising estimations, though they have certain limitations that must be addressed for practical implementation.