Network Optimization Models at Greater Kuparuk Area Using Neural Networks and Genetic Algorithms
Bibliographic record
Abstract
Abstract The Greater Kuparuk Area, located on the North Slope of Alaska, began production in 1981 and has produced over 2.5 billion barrels to date. The area contains six oil fields flowing into three central processing facilities with 47 drillsites and over 1,200 production and injection wells. The facilities are primarily gas constrained due to limits from the gas lift compressors, and secondarily water constrained due to injection pump capacity. An optimization program using the equal slope concept is currently in use for lift gas allocation. A previous attempt to more rigorously optimize the production system using commercial software resulted in better lift gas allocation, but computation time lead to the cessation of its use for daily optimization. The objective of this work was to develop a fast, flexible optimization model that recommends well status, lift gas rates, and water injection rates. The model uses field data and data generated by the previous surface models to develop the hydraulic models as well as current facility conditions and constraints. The model contains four components. The first was a function that estimates producer and injector performance. The second is a function that gathers and interpolates well performance models with physics-based models. Third, the drillsite header pressures were estimated using a neural network. Finally, a genetic algorithm is used to search for the optimal well status, lift gas rate, and water injection rate for each well. Connections were made to databases to run the model using field conditions at any time over the last five years. The hydraulic model for three phase flow utilizes a neural network, whereas a simpler linear based model is used for the water injection system. The hydraulic model was rigorously back tested using field data over a two-year period with weekly model retraining. Drillsite header pressures deviated 6 psi on average from actuals, which is on par with commercial software. The optimization converges in under 90 seconds in a single facility optimization run. The recommendations from the optimization program are expected to increase oil rate 1.5% in the existing system. While production optimization using genetic algorithms and neural networks has been presented for over 20 years, there are not many, if any, known industry applications of optimizing the production and injection networks simultaneously using neural network models. The program was written in Python and deployed on cloud computing. The tool is used to calculate daily net oil benefit per well, prioritize shut-in wells when the facility is constrained, and optimize injection pumps and drillsite configurations. Additionally, the model is designed to accept new engineer-specified source wells to understand the impact of backout when developing new projects. Overall, the model has provided a platform for engineers to make optimization decisions in a complex, interdependent system.
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".