Machine Learning Assisted Framework to Improve Well Rate Estimation Processes for Enhanced Production Surveillance
Bibliographic record
Abstract
Abstract Robust well rate estimation, one of the most important requirements for production surveillance, is essential to understand the performance of a hydrocarbon field and plan optimization activities that can assist in additional production uplift. This paper is concerned with developing a novel framework to combine multiple data sources in order to produce highly reliable well rate estimates. The focus is on improving meter water-cuts (water to total liquid ratio) during measurements using manual spin cuts. Accuracy of in-line automated water-cut meter readings, which measures water-cut during well testing, gets reduced with presence of gas. Water-cuts calculated from spin cut samples, which are manually collected from well test set-up, are more reliable as they do not depend on gas content. Different machine learning modeling approaches were evaluated to estimate a new corrected meter water-cut reading for every well test, using historical spin cut data and closest meter water-cut and temperature readings recorded during collection of these spin cut samples. To account for fluctuations in meter readings resulting from manual calibration of meter, a weighted correction factor ("bias correction") was also added to the improve the predictions. While comparing raw meter reading and model estimates to water cut from spin cut samples, from one year of unseen data, a significant reduction (from 5.6% to 0.5%) in mean error was observed. From evaluations, Tree Bagger models, which are Random Forest models without random selection of features, were found to have the best performance. Moreover, the model was able to capture and respond to sudden shifts in meter readings, possibly originating from manual calibration of the meter. While considering individual wells (or well groups which are tested together through the same meter), poor model performance was observed for few wells where spin cuts consistently recorded values lower than the corresponding meter readings. Model performance was also impacted on well groups where (a) spin cuts taken in close time proximity showed higher variation in water-cuts and (b) spin cut data was sparse. The corrected water-cuts and corresponding phase rates received positive feedback from the operators, underscoring their significance in well rates estimation. This approach uses multiple data sources, real time and historical meter data and, latest and historical manual spin cut data for a surveillance system that runs every few minutes, thus correcting water-cut measurements and coupling it with real time well rate estimations. Additionally, an innovative bias correction method was implemented to improve accuracy of measurements. This framework generates substantial value for upstream businesses with higher well counts (1000+ wells) where data driven approaches are proven to enhance surveillance.
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 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| 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".