Earth, Atmosphere & Environment Sciences
Title : | Development of A Machine Learning Model Based on Experimental Investigations to Quantify Formation Damage Caused by Drilling and Completion Fluids in Indian Oilfields |
Area of research : | Earth, Atmosphere & Environment Sciences |
Focus area : | Formation Damage Analysis |
Principal Investigator : | Dr. Rajat Jain, Indian Institute Of Petroleum And Energy, Visakhapatnam, Andhra Pradesh |
Timeline Start Year : | 2024 |
Timeline End Year : | 2027 |
Contact info : | rajatjain.petro@iipe.ac.in |
Details
Executive Summary : | Formation damage is an important issue with various industries doing business in Exploration and Production of Hydrocarbons. The clogging of pores by solids and polymers is considered to be the primary mechanism responsible for formation damage. Formation damage caused by drilling and completion fluids often happens during the initial stages of production and resulting in a significant reduction in the production rate. This will directly affect the economics of the field project. Hence, customization of drilling and completion fluids needs to be done to minimize formation damage during drilling and completion operations. This will be achieved by carrying filtration studies at High pressure high temperature (HPHT) conditions, rock-fluid interactions and fluid-fluid interaction studies. The design parameters such as fluid rheology, density, pH, temperature, filter cake thickness and removal, filtrate volume, filter cake permeability, and return permeability (for core) will be investigated as per standard procedures. The characterization of rock and fluid samples will be carried out using analytical techniques. Drilling and completion fluid samples will be developed as per field procedures. Rheological and filtration properties using different filter media will be carried out as per API recommended procedures. Further core flow studies with collected core samples and Berea sandstone will be conducted at high pressure and temperature conditions to determine extent of permeability impairment by developed fluid systems. Machine learning (ML) approach will be applied to anticipate the extent of permeability decline in a formation with designed drilling and completion fluids. Complications leading to formation damage can be minimized by developing a model accounting key factors. Such models will consider various factors such as fluid properties, formation properties, and drilling parameters for more accurate predictions. In this proposed research, random-forest, XGBoost, support vector machine, multilayer perception, and multi linear regression will be used for data mining analysis. The regression model's optimal tuning parameter will be assessed using cross-validation and minimizing mean square errors. To conduct the analysis, approximately 70% of the data set points will be utilized to train the models, while the remainder will be used to test model accuracy and evaluate mean absolute error and mean absolute percentage error. The successful outcome of the project will provide a trained model to Indian oil field operators to predict formation damage potential of various fluid systems used in the Indian rock formations without any extra cost and time. |
Co-PI: | Dr. Ranjan Pramanik, Indian Institute Of Petroleum And Energy, Visakhapatnam, Andhra Pradesh-530003 |
Total Budget (INR): | 20,84,240 |
Organizations involved