Development of a Machine Learning -Based Framework for Estimating Rate of Penetration in a Hydrocarbon Reservoir

The primary objective of this research is to enhance the calculation of the Rate of Penetration (ROP) in one of Iran’s oil fields using Machine Learning (ML) techniques. Given the significance of ROP evaluation in optimizing drilling operations and reducing costs, this study aims to develop an accurate and efficient model for predicting ROP based on geological characteristics and drilling parameters. AI methods are employed to improve prediction accuracy and optimize drilling processes and ultimately reduce the time and costs of drilling operations. The data required for model generation were collected from one of the Iranian oil fields, encompassing lithological and formation properties and drilling parameters. The back-propagation multi-layer deep models trained, tested and validated with real data. The results demonstrated that ML techniques can significantly enhance ROP prediction accuracy leading to improved and more controlled drilling processes. Additionally, these models can reduce drilling time and substantially lower operational costs. Two ML methods, Random Forest (RF) and Support Vector Machines (SVM), were utilized for modeling based on available data. Input parameters for the model include Depth In-Out/Meterage (m), Weight on Bit (WOB) Min-Max, Rotations Per Minute (RPM) Min-Max, Gallons Per Minute (GPM) Min-Max, Pump Pressure On-Bottom Min-Max, Mud Weight In Min-Max and Mud Weight Out Min-Max. The model’s output is the calculated ROP with minimal error.

کلیدواژه ها:

Rate of Penetration ، Machine Learning ، Predictive Modeling ، Petroleum Drilling ، Drilling optimization