Accurate and Rapid Forecasts for Geologic Carbon Storage via Learning-Based Inversion-Free Prediction

Dan Lu, Scott L. Painter, Nicholas A. Azzolina, Matthew Burton-Kelly, Tao Jiang, Cody Williamson

Research output: Contribution to journalArticlepeer-review

6 Scopus citations

Abstract

Carbon capture and storage (CCS) is one approach being studied by the U.S. Department of Energy to help mitigate global warming. The process involves capturing CO2 emissions from industrial sources and permanently storing them in deep geologic formations (storage reservoirs). However, CCS projects generally target “green field sites,” where there is often little characterization data and therefore large uncertainty about the petrophysical properties and other geologic attributes of the storage reservoir. Consequently, ensemble-based approaches are often used to forecast multiple realizations prior to CO2 injection to visualize a range of potential outcomes. In addition, monitoring data during injection operations are used to update the pre-injection forecasts and thereby improve agreement between forecasted and observed behavior. Thus, a system for generating accurate, timely forecasts of pressure buildup and CO2 movement and distribution within the storage reservoir and for updating those forecasts via monitoring measurements becomes crucial. This study proposes a learning-based prediction method that can accurately and rapidly forecast spatial distribution of CO2 concentration and pressure with uncertainty quantification without relying on traditional inverse modeling. The machine learning techniques include dimension reduction, multivariate data analysis, and Bayesian learning. The outcome is expected to provide CO2 storage site operators with an effective tool for timely and informative decision making based on limited simulation and monitoring data.

Original languageEnglish
Article number752185
JournalFrontiers in Energy Research
Volume9
DOIs
StatePublished - Jan 12 2022

Funding

Primary funding support for this work is provided by the Science-informed Machine Learning to Accelerate Real Time decision making for Carbon Storage (SMART-CS) Initiative, funded by the US Department of Energy (DOE), Office of Fossil Energy. Additional support is provided by the Artificial Intelligence Initiative as part of the Laboratory Directed Research and Development Program of Oak Ridge National Laboratory, managed by UT-Battelle, LLC, for the US DOE under contract DE-AC05-00OR22725. This research is also sponsored by the Data-Driven Decision Control for Complex To address these challenges, our research aims to develop machine learning (ML) techniques with a potential to provide significant improvements to the conventional history matching-based forecasts, thus enhancing the timeliness and accuracy of information provided to the operator. This paper describes our methods and analyzes their performance in predicting the CO2 plume and pressure distribution in the storage reservoir at a commercial-scale storage project. Our project is part of a large initiative called SMART (Science-informed Machine Learning for Accelerating Real Time Decisions in Subsurface Applications) funded by U.S. Department of Energy with the goal to enable better decisions in CO2 storage operations. Primary funding support for this work is provided by the Science-informed Machine Learning to Accelerate Real Time decision making for Carbon Storage (SMART-CS) Initiative, funded by the US Department of Energy (DOE), Office of Fossil Energy. Additional support is provided by the Artificial Intelligence Initiative as part of the Laboratory Directed Research and Development Program of Oak Ridge National Laboratory, managed by UT-Battelle, LLC, for the US DOE under contract DE-AC05-00OR22725. This research is also sponsored by the Data-Driven Decision Control for Complex Systems (DnC2S) project funded by the US DOE, Office of Advanced Scientific Computing Research.

Keywords

  • Bayesian inference
  • accurate and rapid forecasts
  • carbon capture and storage
  • dimension reduction
  • machine learning
  • saline formations

Fingerprint

Dive into the research topics of 'Accurate and Rapid Forecasts for Geologic Carbon Storage via Learning-Based Inversion-Free Prediction'. Together they form a unique fingerprint.

Cite this