Abstract:A significant volume of anthropogenic carbon dioxide (CO2) is being injected into geological reservoirs at multiple locations throughout the United States. These operations have gained considerable traction in the last decade as a viable option to reduce the long-term environmental footprint of greenhouse gas emissions. However, monitoring the storage reservoir to ensure safe and long-term storage of CO2 for de-risk operations and storage management is undergoing dynamic shifts, expanding opportunities for implementing innovative techniques and applications, especially for the commercial-scale deployment. In this paper, an innovative method comprising multi-tiered analysis has been developed to leverage advanced machine learning (ML) techniques to process passive seismic monitoring data acquired over the two-year injection period along with pumping/injection pressure and rate in the Illinois Basin for potential fracture/fault analysis. Moreover, well log data including imaging logs and core test results are also integrated in the study to add another angle of the understanding for fracture. This study is aimed at contributing to the risk assessment and monitoring design for carbon capture, utilization and storage (CCUS). As a continuation of this effort, we intend to compare the spatial distribution of fractures with the available well logs and core data to deduce a comprehensive understanding of how these fractures correlate with the rock petrophysical properties. The fracture networks obtained from this study would directly benefit field operations and reservoir management decision-making for CCUS, which include the dynamic injection scheduling/workover, risk assessment, and further monitoring design. Furthermore, for ecosystem benefits, the proposed methodology is potentially applicable to oil and gas and renewable assets, such as geothermal development for their fracture-based monitoring and risk reduction as well.

Deep learning based 3D fault detection for carbon capture and storage (CCS)

DeFault: Deep-learning-based Fault Delineation Using the IBDP Passive Seismic Data at the Decatur CO2 Storage Site

Fault Detection and Isolation Methods in Subsea Observation Networks

Uncertainty Quantification Based on Deep-Learning Approach Integrating Time-Lapse Seismic Data for Geological Carbon Storage

Machine Learning Application for CCUS Carbon Storage: Fracture Analysis and Mapping in the Illinois Basin

A Deep Learning Approach to Anomaly Detection in Geological Carbon Sequestration Sites Using Pressure Measurements

3-D Seismic Fault Detection Using Recurrent Convolutional Neural Networks With Compound Loss

Augmented deep learning workflow for robust fault prediction over multiple tectonic regimes

Automatic Seismic Fault Interpretation with the Aid of Data-, Physics-, and Math-assisted Synthetic Data Generation

Seismic Fault Detection with Convolutional Neural Network

Connect the Dots: In Situ 4D Seismic Monitoring of CO2 Storage with Spatio-temporal CNNs

DeepISMNet: three-dimensional implicit structural modeling with convolutional neural network

Data augmentation for 3D seismic fault interpretation using deep learning

Fault Diagnosis of Industrial Process Using Attention Mechanism with 3DCNN-LSTM

MTL-FaultNet: Seismic data reconstruction assisted multi-task deep learning 3D fault interpretation

CO2 storage monitoring based on time-lapse seismic data via deep learning

DeepNRMS: Unsupervised Deep Learning for Noise-Robust CO2 Monitoring in Time-Lapse Seismic Images

3D seismic intelligent prediction of fault-controlled fractured-vuggy reservoirs in carbonate reservoirs based on deep learning method

3D Fault Detection: Using Human Reasoning to Improve Performance of Convolutional Neural Networks

CNN Based Fault Recognition with Multi-Scale Fusion Attention Mechanism