Semester

Fall

Date of Graduation

2025

Document Type

Thesis

Degree Type

MS

College

Statler College of Engineering and Mineral Resources

Committee Chair

Shahab D.Mohaghegh

Committee Co-Chair

Sam Ameri

Committee Member

Kashy Aminian

Abstract

Abstract

Artificial Engineering Intelligence Proxy Modelling for Numerical Reservoir Simulation of CO2 Injection

Mohammad Altahow

This study uses Artificial Intelligence (AI) and Machine Learning (ML) to improve the speed and efficiency of reservoir simulations for CO₂ storage applications. Numerical reservoir simulators, such as CMG, provide accurate predictions but demand extensive computational time and resources, making them impractical for rapid scenario analyses. To tackle this challenge, I developed a Smart Proxy Model using Artificial Neural Networks (ANNs) within the Improve software platform to replicate the results of numerical reservoir simulations conducted with CMG.

The framework predicts key reservoir performance indicators, including pressure distribution and gas saturation evolution. The model was trained and validated using real static and dynamic data from a CO₂ injection field with three injection wells. To capture reservoir variability, 10 unique numerical simulation scenarios were generated, each with a different injection rate profile. Eight scenarios were used for training, while two blind test cases (Run 4 and Run 8) were for validation.

The ANN-based smart proxy model has demonstrated excellent agreement with the blind simulation results, accurately reproducing both the spatial and temporal patterns of reservoir pressure and gas saturation. These outcomes confirm that AI-driven modeling can significantly reduce computational time while maintaining high accuracy. More importantly, this innovative approach empowers us to perform rapid scenario evaluations, quantify uncertainty, and make data-driven decisions in carbon capture and storage (CCS) operations.

Download

Share

COinS