Loading...

Prediction of Foam Injection Performance in Porous Media for EOR using Machine Learning Algorithms

Choopani Rostami, Nader | 2025

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 58408 (06)
  4. University: Sharif University of Technology
  5. Department: Chemical and Petroleum Engineering
  6. Advisor(s): Fatemi, Mobeen
  7. Abstract:
  8. Forecasts indicate that crude oil will supply approximately 20–25% of global energy by 2035. Foam injection, as a novel and effective technique with low sensitivity to gravity and permeability heterogeneity, significantly enhances macroscopic displacement efficiency compared to water, gas, and their alternating injection, and covers a wider portion of the reservoir. Under constraints such as temperature, salinity, or low permeability, foam serves as an efficient alternative for fluid mobility control. This study was designed to investigate and predict the performance of foam injection at the core scale. Parameters such as ultimate oil recovery factor and oil production rate, cumulative gas-oil ratio (GOR), and oil recovery factor (RF) based on injected pore volume were examined. Initially, simulations were conducted using CMG software and validated against published laboratory data. Subsequently, a reliable dataset was prepared for training machine learning models. Decision tree, neural network-augmented gradient boosting, CNN, LightGBM, CatBoost, and XGBoost models were trained for all target variables; however, all of them showed acceptable performance only in predicting the ultimate oil recovery factor, with the best R² value for this variable reaching 0.9921. In predicting RF based on injected pore volume, the decision tree and neural network-augmented gradient boosting yielded the best results. CNN also demonstrated superior performance in predicting cumulative GOR. None of the models successfully predicted oil production rate. The findings of this study assist other researchers in selecting more effective models and avoiding the use of low-performance ones
  9. Keywords:
  10. Foam Injection ; Oil Production Rate (OPR)Curve ; Machine Learning ; Artificial Intelligence ; Foam Injection Performance Prediction ; Oil Production Rate Prediction ; Cumulative Gas-Oil Ratio (GOR) Prediction ; Oil Recovery Factor Prediction

 Digital Object List

 Bookmark

No TOC