Field Test Results for Real-time ROP Optimization Using Machine Learning and Downhole Vibration Monitoring - A Case Study
Bibliographic record
Abstract
Abstract A case study for a real-time field test of a machine learning (ML) ROP prediction and optimization algorithm and a vendor-neutral vibration monitoring system is presented for ten wells in Northeastern British Columbia, Canada. A novel auto-calibration feature adjusts the ML model in real-time to account for prediction bias. The paper is of interest to operators and service companies seeking to accelerate uptake of Artificial Intelligence (AI) by rig personnel. A ten-well campaign in three target formations was drilled from one rig in the lateral sections to test the ML ROP prediction/optimization system. During the first two laterals, the operator office engineers visited the rig to train the rig team and gain their buy-in. For the remaining wells, tests were run in real-time advisory mode. Field test objectives were to develop trust in the ML model, validate real-time vibration monitoring tool with real-time downhole vibration data, and obtain feedback from the rig and office on functionality to accelerate uptake of AI. Initially, the ML ROP model passed all success criteria in two of three formations, or four of six wells. One formation failed the ROP accuracy criterion because the predicted ROP was consistently too high, but the ML model accurately captured the variance. This led to the development of a novel automated calibration procedure that adjusts the "bias" of the machine learning ROP prediction in a manner like calibrating physics-based models (such as torque and drag hookload) using a calibration factor calculated in real-time by comparing predictions with actual ROP values. This enhancement enables meeting accuracy acceptance criteria and has opened the door for a broader application of the method in other formations and basins. To date the model has been successfully deployed for the lateral section and for bottomhole assemblies (BHAs) with a positive displacement motor. The vibration monitoring system successfully provided real-time data to the operator at the rig and office that is generally available only to the MWD provider in real time, which enables the operator to gain new insights for situational awareness and decision making. Feedback from rig personnel was very valuable and included new functionality, such as their ability to change parameter limits in the ROP system, which has been implemented and successfully used by the operator.
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 | 0.002 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.000 |
| 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".