Understanding the Controlling Factors for CO₂ Sequestration in Depleted Shale Reservoirs Using Data Analytics and...

Area of Science:

• Geosciences
• Petroleum Engineering
• Environmental Science

Background:

• Carbon capture and sequestration (CCS) is vital for reducing industrial CO2 emissions.
• Unconventional reservoirs, like shale formations, offer significant potential for CO2 storage.
• Understanding CO2 sequestration dynamics in these reservoirs is crucial for developing effective CCS strategies.

Purpose of the Study:

• To identify key parameters influencing CO2 sequestration in depleted shale reservoirs.
• To develop and evaluate machine learning models for predicting CO2 injection volumes.
• To provide insights for optimizing CCS operations in unconventional formations.

Main Methods:

• Utilized an extensive dataset of numerical simulation results.
• Applied data analytics and various machine learning algorithms (multiple linear regression, regression tree, bagging, random forest, gradient boosting).
• Performed variable importance analysis to rank influential reservoir and operational parameters.

Main Results:

• Random forest models demonstrated the highest predictive accuracy for cumulative CO2 injected.
• Regression tree models showed the poorest performance due to overfitting.
• Stimulated reservoir volume fracture permeability was identified as the most significant predictor of CO2 sequestration.

Conclusions:

• Machine learning, particularly random forest, is effective for predicting CO2 sequestration performance in shale reservoirs.
• Fracture permeability within the stimulated reservoir volume is a primary control on CO2 storage capacity.
• The study provides valuable insights for the oil and gas industry regarding CO2 sequestration in depleted shale formations.