A novel reservoir simulation model based on physics informed neural networks
Bibliographic record
Abstract
Surrogate models are widely used for reservoir simulations in the petroleum industry to improve computational efficiency. However, the traditional surrogate model mainly relies on the data collected from production wells (e.g., well bottom pressure data and well production data) and ignores the physical mechanism of underground fluid flow; therefore, the surrogate model will be invalid in the case of insufficient data samples. In response to these challenges, a Hard-Soft physics informed neural network (HS-PINN) was proposed to simulate pressure fluctuations around producing wells without relying on any labeled data, where two coupled fully connected neural networks were comprised to control the Hard and Soft constraint conditions. Specifically, in the “Soft Constraint” condition, we employ a modified Lorentz function to incorporate underground flow theory and permeability fields into the loss function. Meanwhile, in the “Hard Constraint” condition, we incorporate an enforcement function in the “output layer” to ensure the network outputs satisfy the boundary and initial conditions. To demonstrate the HS-PINN model's robustness and accuracy abilities, we tested it for single and multi-well production in both noisy low-fidelity and high-fidelity geologic reservoir environments, and the HS-PINN prediction errors were less than 1% in both cases compared to simulation results by the commercial software “COMSOL.” Additionally, we assessed the impacts of varying well interference intensities, adjustments in collocation points counts within the control equations, and diverse geological characteristics on model performance to validate the generalization and stability of HS-PINN. Moreover, the HS-PINN-based surrogate model significantly improves the efficiency of uncertainty quantification tasks compared to simulation-based approaches, requiring only 8% of the computational time. The deep-learning surrogate models developed in this work offer a novel and efficient approach for simulating reservoir development.
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 | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.000 | 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) | 0.002 | 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".