A machine‐learning framework to estimate saturation changes from 4D seismic data using reservoir models
Bibliographic record
Abstract
Abstract Time‐lapse seismic (four‐dimensional seismic) data play a preeminent role in closed‐loop reservoir management by providing a full‐field image of dynamic reservoir behaviour during production. Nonetheless, the multidisciplinary nature of four‐dimensional closed‐loop approaches demands more quantitative and fast methods to integrate rock physics models, reservoir flow simulation models and four‐dimensional seismic analysis. In this work, we tackle this time‐consuming and expensive process and develop a data‐driven quantitative approach to leverage the inherent physics between four‐dimensional seismic and reservoir property changes. We propose an inversion flow method using machine learning strategies to estimate changes in reservoir properties directly from a fast‐derived four‐dimensional seismic attribute. This study was carried out in a Brazilian deep‐water field where production started in 2013 with 3 years of production and injection history. For this reservoir, the estimation of fluid saturation maps is a critical objective to assist engineers with data assimilation procedures. We also generated millions of training data samples using 200 simulation models from the field mentioned above (before history matching) to highlight the benefits of restricting training samples to proper fluid flow consistent combinations. Results demonstrate high prediction accuracy for the targeted reservoir property changes. Additionally, it provides insights for the detection of sweet spots and positioning of infill wells. We significantly simplify the four‐dimensional seismic integration process, allowing initial engagements of reservoir engineers with decision‐making processes and data assimilation applications.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".