Data-Driven Steam Optimization for SAGD
Bibliographic record
Abstract
Abstract Since decades, steam-assisted oil recovery processes have been successfully deployed in heavy oil reservoirs to extract bitumen/heavy oil. Current resource allocation practices mostly involve reservoir model-based open loop optimization at the planning stage and its periodic recurrence. However, such decades-old strategies need a complete overhaul as they ignore dynamic changes in reservoir conditions and surface facilities, ultimately rendering heavy oil production economically unsustainable in the low-oil-price environment. Since steam supply costs account for more than 50% of total operating costs, a data-driven strategy that transforms the data available from various sensors into meaningful steam allocation decisions requires further attention. In this research, we propose a purely data-driven algorithm that maximizes the economic objective function by allocating an optimal amount of steam to different well pads. The method primarily constitutes two components: forecasting and nonlinear optimization. A dynamic model is used to relate different variables in historical field data that were measured at regular time intervals and can be used to compute economic performance indicators (EPI). The variables in the model are cumulative in nature since they can represent the temporal changes in reservoir conditions. Accurate prediction of EPI is ensured by retraining the regression model using the latest available data. Then, predicted EPI is optimized using a nonlinear optimization algorithm subject to amplitude and rate saturation constraints on decision variables i.e., the amount of steam allocated to each well pad. The proposed steam allocation strategy is tested on 2 well pads (each containing 10 wells) of an oil sands reservoir located near Fort McMurray in Alberta, Canada. After an exploratory analysis of production history, an output error (OE) model is built between logarithmically transformed cumulative steam injection and cumulative oil production for each well pad. Commonly used net-present-value (NPV) is considered as EPI to be maximized. Optimization of the objective function is subject to distinct operating conditions and realistic constraints. By comparing results with field production history, it can be observed that optimum steam injection profiles for both well pads are significantly different than that of a field. In fact, the proposed algorithm provides smooth and consistent steam injection rates, unlike field injection history. Also, the lower steam-oil ratio is achieved for both well pads, ultimately translating into ∼19 % higher NPV when compared with field data. Inspired from state-of-the-art control techniques, the proposed steam allocation algorithm provides a generic data-driven framework that can consider any number of well pads, EPIs, and amount of past data. It is computationally inexpensive as no numerical simulations are required. Overall, it can potentially reduce the energy required to extract heavy oil and increase the revenue while inflicting no additional capital cost and reducing greenhouse gas emissions.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 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 teacher head, 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".