Multimodal Auto-regressive Transformer Surrogate for Modeling Variable Operations and Quantifying Uncertainty in Geological Carbon Storage
Researchers developed a new multimodal auto-regressive transformer surrogate to model variable operations and quantify uncertainty in geological carbon storage. The model processes three input modalities through separate encoders and fuses them via self-attention in a tra…
Intelligence analysis by Llama
The surrogate model is trained to predict saturation and pressure at monitoring locations, total injected and mobile CO2 mass, and saturation footprints. It achieves a median saturation MAE of 0.028 and median relative errors of 0.2-5% for the other quantities of interest.
Imagine you're trying to store carbon dioxide underground. It's like a big puzzle, and you need to figure out the best way to do it. Researchers created a special computer model that can help solve this puzzle by looking at different pieces of information and using them to make predictions. This model is like a super-smart assistant that can help us store carbon dioxide more efficiently and reduce uncertainty in the process.
Analysis
A Multimodal Approach to Modeling Variable Operations
The use of variable well perforation and injection strategies can improve the efficiency of geological carbon storage operations. To model these operations under geological uncertainty, researchers developed a new multimodal auto-regressive transformer surrogate. This model processes three input modalities - the 3D geomodel, scalar parameters characterizing relative permeability functions, and control variables - through separate encoders. These are fused via self-attention in a transformer encoder, and a temporal decoder generates predictions auto-regressively through encoder-decoder cross-attention.
Training the Surrogate Model
The surrogate model is trained using 4000 GEOS flow simulations to predict saturation and pressure at monitoring locations, total injected and mobile CO2 mass, and saturation footprints. For a new test set, involving randomly sampled geomodels and control variables, the surrogate achieves a median saturation MAE of 0.028 and median relative errors of 0.2-5% for the other quantities of interest. Importantly, it captures the switch from rate to bottom-hole-pressure control.
Uncertainty Reduction through Data Assimilation
The surrogate model is used within a hierarchical Markov chain Monte Carlo data assimilation procedure for a synthetic true model under three operational strategies. Substantial uncertainty reduction is achieved for key metaparameters, particularly the fault permeabilities. Posterior predictions for saturation footprints and total injected and mobile CO2 mass are also shown to be generally consistent with true model results.
Key points
- Researchers developed a new multimodal auto-regressive transformer surrogate to model variable operations and quantify uncertainty in geological carbon storage.
- The model processes three input modalities through separate encoders and fuses them via self-attention in a transformer encoder.
- The surrogate model is trained to predict saturation and pressure at monitoring locations, total injected and mobile CO2 mass, and saturation footprints.
- The model achieves a median saturation MAE of 0.028 and median relative errors of 0.2-5% for the other quantities of interest.
- The surrogate model is used within a hierarchical Markov chain Monte Carlo data assimilation procedure for a synthetic true model under three operational strategies.
If this development plays out positively, it could lead to more efficient and effective carbon storage operations. This, in turn, could help reduce greenhouse gas emissions and mitigate the effects of climate change.
However, there are also potential risks associated with this development. For example, if the model is not accurate or is not properly validated, it could lead to incorrect predictions and potentially harmful decisions. Additionally, the increased use of carbon storage operations could lead to new environmental concerns, such as the potential for leaks or other accidents.
