Hybrid CFD and machine learning analysis of CO2 enhanced oil recovery in naturally fractured reservoirs
Гибридный анализ повышения нефтеотдачи с применением CO2 в естественнотрещиноватых коллекторах на основе вычислительной гидродинамики и машинного обучения
2026-01-07
SCID: 54.1/rtkdbbf8
Discuss with AI
CO2 enhanced oil recoveryGaussian process regressionXGBoostcomputational fluid dynamicsnaturally fractured reservoirs
Figures from the paper
Abstract (AI)
CO2 based Enhanced Oil Recovery (EOR) in unconventional reservoirs is an emerging technology. Scientific research efforts are directed towards understanding the propagation of CO2 front due to the complex interplay between CO2 injection and saturation, and reservoir’s constitutive relationships. Conventional methods for characterising CO2-EOR rely on high-fidelity numerical solutions that often result in over or under prediction of CO2 geosequestration. In this study, we develop a novel hybrid Computational Fluid Dynamics (CFD) and Machine Learning (ML) framework that allows for rapid CO2 geosequestration prediction and its optimal injection. Very low or high injection rates have been shown to result in low sweep efficiency or excessive entry pressure, while an intermediate injection rate offers the best balance between the two. CFD data-driven Gaussian Process Regression (GPR) and Extreme Gradient Boosting (XGBoost) models have been developed, trained and tested for predicting CO2 saturation in the reservoir. Comparative analysis indicates that GPR outperforms XGBoost in terms of its predictive performance and robustness. Through the analysis of layer-resolved CO2 front displacement and development of data-driven surrogate models, this study contributes a novel framework for CO2-EOR predictive modelling and optimising injection strategies in naturally fractured reservoirs.
Key Findings
1
A hybrid CFD–machine learning framework was developed for rapid prediction of CO2 geosequestration and optimization of injection strategies in naturally fractured reservoirs.
2
Gaussian Process Regression and Extreme Gradient Boosting models were trained on CFD data to predict reservoir CO2 saturation.
3
Gaussian Process Regression outperformed XGBoost in predictive performance and robustness for CO2 saturation forecasting.
4
Layer-resolved CO2 front analysis and data-driven surrogate modeling support improved predictive modeling and injection optimization for CO2-EOR.
5
Very low injection rates produce poor sweep efficiency, whereas very high rates cause excessive entry pressure; intermediate rates provide the best balance.
Research Object
CO2 enhanced oil recovery and geosequestration in naturally fractured reservoirs
Research Subject
CO2 front propagation, saturation prediction, sweep efficiency, and injection-rate optimization
Publication Details
Publication Date
2026-01-07
Journal
Publisher
ISSN
Cited by
4
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest