Summary of Accelerated Training Of Deep Learning Surrogate Models For Surface Displacement and Flow, with Application to Mcmc-based History Matching Of Co2 Storage Operations, by Yifu Han et al.
Accelerated training of deep learning surrogate models for surface displacement and flow, with application to MCMC-based history matching of CO2 storage operations
by Yifu Han, Francois P. Hamon, Louis J. Durlofsky
First submitted to arxiv on: 20 Aug 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The new surrogate modeling framework introduced in this paper aims to predict CO2 saturation, pressure, and surface displacement for carbon storage operations, which requires efficient training due to computational demands. The approach combines inexpensive flow-only simulations with a smaller number of expensive coupled flow-geomechanics runs, leveraging effective rock compressibility for accurate predictions. Recurrent residual U-Network architectures are applied for saturation and pressure surrogate models, while a new model is introduced for surface displacement, which accepts geomodel quantities and surrogate predictions as inputs. The median relative error for the test set remains below 4% for all variables. The surrogate models are incorporated into a hierarchical Markov chain Monte Carlo history matching workflow, considering high prior uncertainty and characterizing geological scenario parameters (metaparameters) and realizations. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper presents a new way to predict how carbon dioxide moves underground, which is important for storing CO2 safely. Instead of running many expensive simulations, the method combines cheaper flow-only simulations with fewer but more complex simulations that include geomechanics. This helps make predictions accurate while saving time and resources. The approach uses special computer models to learn from data and make better guesses about what’s happening underground. By combining data from monitoring wells and surface sensors, the method can improve its predictions and reduce uncertainty. |
