Author ORCID Identifier

https://orcid.org/0000-0001-6843-7557

Semester

Summer

Date of Graduation

2024

Document Type

Dissertation

Degree Type

PhD

College

Statler College of Engineering and Mineral Resources

Department

Petroleum and Natural Gas Engineering

Committee Chair

Shahab Mohaghegh

Committee Co-Chair

Samuel Ameri

Committee Member

Samuel Ameri

Committee Member

Kashy Aminian

Committee Member

Kirk Wilhelmsen

Committee Member

Mohammed El Sgher

Abstract

The application of numerical reservoir simulation (NRS) has been a common approach within the oil and gas industry for decades, providing a means to model and forecast dynamic subsurface interactions, as a basis for reservoir management and development decisions. These techniques have expanded to application within carbon capture utilization and storage (CCUS) projects as domestic and global policy shift towards reducing carbon emissions while maintaining the energy needs of our modern society. NRS techniques have become a core process for permitting approval in Class VI (large-scale geological sequestration) wells due to the fundamental similarity of these types of subsurface processes. This has provided the energy industry an opportunity to apply its expertise in modeling subsurface phenomena from hydrocarbon production to geological sequestration problems, predominantly focused on the injection and storage of carbon dioxide (CO2).

While NRS models are based upon our fundamental understanding of math, physics, and engineering; sparsity of measurement often makes them inherently subjective, requiring stochastic modeling techniques to provide quantifiable uncertainty for management decisions. In geological sequestration, this problem is exacerbated as (1) often the number of well locations is limited to one or two places, (2) subsurface response, which guides model refinement, is yet to be seen as injection has not taken place, (3) interest of long-term storage integrity leads to much longer periods of investigation, where modeling decisions may introduce larger uncertainties to these long-term forecasts.

Although not an all-encompassing list, it is meant to highlight the added uncertainty by the nature of these problems. This compounds the feedback-loop already present in traditional NRS studies, as added complexity and uncertainty require additional models, leading to a lack of computational resources. Traditional proxy modeling techniques attempt to combat this resource intensive process by simplification of the underlying physics or grid domain, but this comes at the cost of added approximation and uncertainty.

This work applies the use of artificial intelligence (AI) and machine learning (ML) from a unique, engineering-based perspective. This is accomplished through a series of techniques, commonly referred to as deep-learning, starting with a full-field AI representation of the reservoir, or Smart Proxy Model, which is an ensemble of various AI-ML algorithms to develop an accurate representation of subsurface interaction. The use of a real dataset from the Illinois Basin Decatur Project (IBDP) presents unique challenges in translating complexity, which is highlighted throughout the model development and feature generation process.

The smart proxy results show the capability of providing accurate, full-fidelity results of subsurface interactions in seconds. This is carried into testing the viability of utilizing the smart proxy model, where the speed of a combined smart proxy optimization ensemble is of interest. These results display that the approach is capable of handling this coupled approach, with iterations typically under one second for feature updates, deployment and updating relevant inputs, allowing for additional insight by exploiting the fast-predictive nature of AI-ML algorithms. This approach offers a significant advantage over traditional methods by maintaining high accuracy while reducing computational time and resources. This provides a pathway for more effective reservoir management and decision-making processes in both hydrocarbon production and CO2 sequestration projects.

Download

Share

COinS